Gaurav S. Sukhatme

VLDB 2026 Research / reviewers in the wild / expert

dblp:s/GauravSSukhatme · also G. Stefano Sukhatme · DBLP profile ↗

← Back

244ranked-venue papers

2as first author

36since 2021 · last 2025

0000-0003-2408-474XORCID · verified

Domains — the database's venue-derived domains; a paper can count in several

Artificial intelligence and machine learning · 194 · 2 first-author · 29 since 2021Systems, architecture and hardware · 174 · 2 first-author · 20 since 2021Applied, interdisciplinary, general and emerging computing · 24 · 5 since 2021Computer networks · 19 · 2 since 2021Graphics, computer vision, multimedia, augmented reality and games · 3 · 1 since 2021Software engineering, systems software and programming languages · 1

Year	Publication	Venue	Position
2025	Zero Shot Generalization of Vision-Based RL Without Data Augmentation abstract Generalizing vision-based reinforcement learning (RL) agents to novel environments remains a difficult and open challenge. Current trends are to collect large-scale datasets or use data augmentation techniques to prevent overfitting and improve downstream generalization. However, the computational and data collection costs increase exponentially with the number of task variations and can destabilize the already difficult task of training RL agents. In this work, we take inspiration from recent advances in computational neuroscience and propose a model, Associative Latent DisentAnglement (ALDA), that builds on standard off-policy RL towards zero-shot generalization. Specifically, we revisit the role of latent disentanglement in RL and show how combining it with a model of associative memory achieves zero-shot generalization on difficult task variations without relying on data augmentation. Finally, we formally show that data augmentation techniques are a form of weak disentanglement and discuss the implications of this insight. Sumeet Batra, Gaurav S. Sukhatme	ICML	2
2025	SAFE-GIL: SAFEty Guided Imitation Learning for Robotic Systems abstract Behavior cloning (BC) is a widely used approach in imitation learning where a robot learns a control policy by observing an expert supervisor. However the learned policy can make errors and might lead to safety violations which limits their utility in safety-critical robotics applications. While prior works have tried improving a BC policy via additional real or synthetic action labels adversarial training or runtime filtering none of them explicitly focus on reducing the BC policy's safety violations during training time. We propose SAFE-GIL a design-time method to learn safety-aware behavior cloning policies. SAFE-GIL deliberately injects adversarial disturbance in the system during data collection to guide the expert towards safety-critical states. This disturbance injection simulates potential policy errors that the system might encounter during the test time. By ensuring that training more closely replicates expert behavior in safety-critical states our approach results in safer policies despite policy errors during the test time. We further develop a reachability-based method to compute this adversarial disturbance. We compare SAFE-GIL with various behavior cloning techniques and online safety-filtering methods in three domains autonomous ground navigation aircraft taxiing and aerial navigation on a quadrotor testbed. Our method demonstrates a significant reduction in safety failures particularly in low data regimes where the likelihood of learning errors and therefore safety violations is higher. See our website here: https://y-u-c.github.io/safegil/. Yusuf Umut Ciftci, Darren Chiu, Zeyuan Feng, Gaurav S. Sukhatme, Somil Bansal	ICRA	4
2025	Resilient Multi-Robot Target Tracking with Sensing and Communication Danger Zones abstract Multi-robot collaboration for target tracking in adversarial environments poses significant challenges, including system failures, dynamic priority shifts, and other unpredictable factors. These challenges become even more pronounced when the environment is unknown. In this paper, we propose a resilient coordination framework for multi-robot, multi-target tracking in environments with unknown sensing and communication danger zones. We consider scenarios where failures caused by these danger zones are probabilistic and temporary, allowing robots to escape from danger zones to minimize the risk of future failures. We formulate this problem as a nonlinear optimization with soft chance constraints, enabling real-time adjustments to robot behaviors based on varying types of dangers and failures. This approach dynamically balances target tracking performance and resilience, adapting to evolving sensing and communication conditions in real-time. To validate the effectiveness of the proposed method, we assess its performance across various tracking scenarios, benchmark it against methods without resilient adaptation and collaboration, and conduct several real-world experiments. Peihan Li, Yuwei Wu 0005, Gaurav S. Sukhatme, Vijay Kumar 0001, Lifeng Zhou 0001	IROS	4
2025	Toward Efficient MPPI Trajectory Generation With Unscented Guidance: U-MPPI Control Strategy abstract The classical Model Predictive Path Integral (MPPI) control framework, while effective in many applications, lacks reliable safety features since it relies due to its reliance on arisk-neutraltrajectory evaluation technique, which can present challenges for safety-critical applications such as autonomous driving. Furthermore, if when the majority of MPPI sampled trajectories concentrate in high-cost regions, it may generate aninfeasiblecontrol sequence. To address this challenge, we propose the U-MPPI control strategy, a novel methodology that can effectively manage system uncertainties while integrating a more efficient trajectory sampling strategy. The core concept is to leverage the Unscented Transform (UT) to propagate not only the mean but also the covariance of the system dynamics, going beyond the traditional MPPI method. As a result, it introduces a novel and more efficient trajectory sampling strategy, significantly enhancing state-space exploration and ultimately reducing the risk of being trapped in local minima. Furthermore, by leveraging the uncertainty information provided by UT, we incorporate arisk-sensitivecost function that explicitly accounts for risk or uncertainty throughout the trajectory evaluation process, resulting in a more resilient control system capable of handling uncertain conditions. By conducting extensive simulations of 2D aggressive autonomous navigation in both known and unknown cluttered environments, we verify the efficiency and robustness of our proposed U-MPPI control strategy compared to the baseline MPPI. We further validate the practicality of U-MPPI through real-world demonstrations in unknown cluttered environments, showcasing its superior ability to incorporate both the UT and local costmap into the optimization problem without introducing additional complexity. Ihab S. Mohamed, Junhong Xu, Gaurav S. Sukhatme, Lantao Liu	IEEE Trans. Robotics	3
2025	DREAM: Decentralized Real-Time Asynchronous Probabilistic Trajectory Planning for Collision-Free Multirobot Navigation in Cluttered Environments abstract Collision-free navigation in cluttered environments with static and dynamic obstacles is essential for many multirobot tasks. Dynamic obstacles may also be interactive, i.e., their behavior varies based on the behavior of other entities. We propose a novel representation for interactive behavior of dynamic obstacles and a decentralized real-time multirobot trajectory planning algorithm allowing interrobot collision avoidance as well as static and dynamic obstacle avoidance. Our planner simulates the behavior of dynamic obstacles, accounting for interactivity. We account for the perception inaccuracy of static and prediction inaccuracy of dynamic obstacles. We handle asynchronous planning between teammates and message delays, drops, and reorderings. We evaluate our algorithm in simulations using 25400 random cases and compare it against three state-of-the-art baselines using 2100 random cases. Our algorithm achieves up to 1.68× success rate using as low as 0.28× time in single-robot, and up to 2.15× success rate using as low as 0.36× time in multirobot cases compared to the best baseline. We implement our planner on real quadrotors to show its real-world applicability. Baskin Senbaslar, Gaurav S. Sukhatme	IEEE Trans. Robotics	2
2024	VLN-Video: Utilizing Driving Videos for Outdoor Vision-and-Language Navigation abstract Outdoor Vision-and-Language Navigation (VLN) requires an agent to navigate through realistic 3D outdoor environments based on natural language instructions. The performance of existing VLN methods is limited by insufficient diversity in navigation environments and limited training data. To address these issues, we propose VLN-Video, which utilizes the diverse outdoor environments present in driving videos in multiple cities in the U.S. augmented with automatically generated navigation instructions and actions to improve outdoor VLN performance. VLN-Video combines the best of intuitive classical approaches and modern deep learning techniques, using template infilling to generate grounded non-repetitive navigation instructions, combined with an image rotation similarity based navigation action predictor to obtain VLN style data from driving videos for pretraining deep learning VLN models. We pre-train the model on the Touchdown dataset and our video-augmented dataset created from driving videos with three proxy tasks: Masked Language Modeling, Instruction and Trajectory Matching, and Next Action Prediction, so as to learn temporally-aware and visually-aligned instruction representations. The learned instruction representation is adapted to the state-of-the-art navigation agent when fine-tuning on the Touchdown dataset. Empirical results demonstrate that VLN-Video significantly outperforms previous state-of-the-art models by 2.1% in task completion rate, achieving a new state-of-the-art on the Touchdown dataset. Jialu Li 0001, Aishwarya Padmakumar, Gaurav S. Sukhatme, Mohit Bansal	AAAI	3
2024	Proximal Policy Gradient Arborescence for Quality Diversity Reinforcement Learning abstract Training generally capable agents that thoroughly explore their environment and learn new and diverse skills is a long-term goal of robot learning. Quality Diversity Reinforcement Learning (QD-RL) is an emerging research area that blends the best aspects of both fields – Quality Diversity (QD) provides a principled form of exploration and produces collections of behaviorally diverse agents, while Reinforcement Learning (RL) provides a powerful performance improvement operator enabling generalization across tasks and dynamic environments. Existing QD-RL approaches have been constrained to sample efficient, deterministic off- policy RL algorithms and/or evolution strategies and struggle with highly stochastic environments. In this work, we, for the first time, adapt on-policy RL, specifically Proximal Policy Optimization (PPO), to the Differentiable Quality Diversity (DQD) framework and propose several changes that enable efficient optimization and discovery of novel skills on high-dimensional, stochastic robotics tasks. Our new algorithm, Proximal Policy Gradient Arborescence (PPGA), achieves state-of- the-art results, including a 4x improvement in best reward over baselines on the challenging humanoid domain. Sumeet Batra, Bryon Tjanaka, Matthew C. Fontaine, Aleksei Petrenko, Stefanos Nikolaidis, Gaurav S. Sukhatme	ICLR	6
2024	HyperPPO: A scalable method for finding small policies for robotic control abstract Models with fewer parameters are necessary for the neural control of memory-limited, performant robots. Finding these smaller neural network architectures can be time-consuming. We propose HyperPPO, an on-policy reinforcement learning algorithm that utilizes graph hypernetworks to estimate the weights of multiple neural architectures simultaneously. Our method estimates weights for networks that are much smaller than those in common-use networks yet encode highly performant policies. We obtain multiple trained policies at the same time while maintaining sample efficiency and provide the user the choice of picking a network architecture that satisfies their computational constraints. We show that our method scales well - more training resources produce faster convergence to higher-performing architectures. We demonstrate that the neural policies estimated by HyperPPO are capable of decentralized control of a Crazyflie2.1 quadrotor. Website: https://sites.google.com/usc.edu/hyperppo Shashank Hegde, Zhehui Huang, Gaurav S. Sukhatme	ICRA	3
2024	Collision Avoidance and Navigation for a Quadrotor Swarm Using End-to-end Deep Reinforcement Learning abstract End-to-end deep reinforcement learning (DRL) for quadrotor control promises many benefits – easy deployment, task generalization and real-time execution capability. Prior end-to-end DRL-based methods have showcased the ability to deploy learned controllers onto single quadrotors or quadrotor teams maneuvering in simple, obstacle-free environments. However, the addition of obstacles increases the number of possible interactions exponentially, thereby increasing the difficulty of training RL policies. In this work, we propose an end-to-end DRL approach to control quadrotor swarms in environments with obstacles. We provide our agents a curriculum and a replay buffer of the clipped collision episodes to improve performance in obstacle-rich environments. We implement an attention mechanism to attend to the neighbor robots and obstacle interactions - the first successful demonstration of this mechanism on policies for swarm behavior deployed on severely compute-constrained hardware. Our work is the first work that demonstrates the possibility of learning neighbor-avoiding and obstacle-avoiding control policies trained with end-to-end DRL that transfers zero-shot to real quadrotors. Our approach scales to 32 robots with 80% obstacle density in simulation and 8 robots with 20% obstacle density in physical deployment. Website: https://sites.google.com/view/obst-avoid-swarm-rl Zhehui Huang, Zhaojing Yang, Rahul Krupani, Baskin Senbaslar, Sumeet Batra, Gaurav S. Sukhatme	ICRA	6
2024	CppFlow: Generative Inverse Kinematics for Efficient and Robust Cartesian Path Planning abstract In this work we present CppFlow - a novel and performant planner for the Cartesian Path Planning problem, which finds valid trajectories up to 129x faster than current methods, while also succeeding on more difficult problems where others fail. At the core of the proposed algorithm is the use of a learned, generative Inverse Kinematics solver, which is able to efficiently produce promising entire candidate solution trajectories on the GPU. Precise, valid solutions are then found through classical approaches such as differentiable programming, global search, and optimization. In combining approaches from these two paradigms we get the best of both worlds - efficient approximate solutions from generative AI which are made exact using the guarantees of traditional planning and optimization. We evaluate our system against other state of the art methods on a set of established baselines as well as new ones introduced in this work and find that our method significantly outperforms others in terms of the time to find a valid solution and planning success rate, and performs comparably in terms of trajectory length over time. Additional results and an open source implementation is available at https://jstmn.github.io/cppflow-website/. Jeremy Morgan, David Millard 0001, Gaurav S. Sukhatme	ICRA	3
2024	Open X-Embodiment: Robotic Learning Datasets and RT-X Models : Open X-Embodiment Collaboration abstract Large, high-capacity models trained on diverse datasets have shown remarkable successes on efficiently tackling downstream applications. In domains from NLP to Computer Vision, this has led to a consolidation of pretrained models, with general pretrained backbones serving as a starting point for many applications. Can such a consolidation happen in robotics? Conventionally, robotic learning methods train a separate model for every application, every robot, and even every environment. Can we instead train "generalist" X-robot policy that can be adapted efficiently to new robots, tasks, and environments? In this paper, we provide datasets in standardized data formats and models to make it possible to explore this possibility in the context of robotic manipulation, alongside experimental results that provide an example of effective X-robot policies. We assemble a dataset from 22 different robots collected through a collaboration between 21 institutions, demonstrating 527 skills (160266 tasks). We show that a high-capacity model trained on this data, which we call RT-X, exhibits positive transfer and improves the capabilities of multiple robots by leveraging experience from other platforms. The project website is robotics-transformer-x.github.io. Abigail O'Neill, Abhiram Maddukuri, Abhishek Gupta 0004, Abhishek Padalkar, Abraham Lee, Acorn Pooley, Agrim Gupta, Ajay Mandlekar, Ajinkya Jain, Albert Tung, Alex Bewley, Alex Irpan, Alexander Khazatsky, Anant Rai, Anchit Gupta, Andrew E. Wang, Anikait Singh, Animesh Garg, Aniruddha Kembhavi, Annie Xie, Anthony Brohan, Antonin Raffin, Archit Sharma, Arefeh Yavary, Arhan Jain, Ashwin Balakrishna, Ayzaan Wahid, Ben Burgess-Limerick, Bernhard Schölkopf, Blake Wulfe, Brian Ichter, Cewu Lu, Charles Xu 0003, Charlotte Le, Chelsea Finn, Chen Wang 0053, Chenfeng Xu, Cheng Chi 0001, Chenguang Huang, Christine Chan, Christopher Agia, Chuer Pan, Chuyuan Fu, Coline Devin, Danfei Xu, Daniel Morton, Danny Drieß, Daphne Chen, Deepak Pathak, Dhruv Shah, Dieter Büchler, Dinesh Jayaraman, Dmitry Kalashnikov, Dorsa Sadigh, Edward Johns, Ethan Paul Foster, Fangchen Liu, Federico Ceola, Fei Xia 0002, Feiyu Zhao, Freek Stulp, Gaoyue Zhou, Gaurav S. Sukhatme, Gautam Salhotra, Gilbert Feng, Giulio Schiavi, Glen Berseth, Gregory Kahn, Guanzhi Wang, Hao Su 0001, Haoshu Fang, Henghui Bao, Heni Ben Amor, Henrik I. Christensen, Hiroki Furuta, Homer Walke, Hongjie Fang, Huy Ha, Igor Mordatch, Ilija Radosavovic, Isabel Leal, Jacky Liang, Jad Abou-Chakra, Jaehyung Kim 0001, Jaimyn Drake, Jan Peters 0001, Jan Schneider 0007, Jasmine Hsu, Jeannette Bohg, Jeffrey T. Bingham, Jensen Gao, Jiaheng Hu, Jiajun Wu 0001, Jiankai Sun, Jianlan Luo, Jiayuan Gu, Jie Tan 0001, Jihoon Oh, Jimmy Wu, Jingpei Lu, Jitendra Malik, João Silvério, Joey Hejna, Jonathan Booher, Jonathan Tompson, Jonathan Yang, Jordi Salvador, Joseph J. Lim, Junhyek Han, Kanishka Rao, Karl Pertsch, Karol Hausman, Keegan Go, Keerthana Gopalakrishnan, Kenneth Y. Goldberg, Kendra Byrne, Kenneth Oslund, Kento Kawaharazuka, Kevin Black, Kevin Zhang 0002, Kiana Ehsani, Kiran Lekkala, Kirsty Ellis, Krishan Rana, Krishnan Srinivasan, Kuan Fang, Kunal Pratap Singh, Kuo-Hao Zeng, Kyle Hatch, Kyle Hsu, Laurent Itti, Yunliang Chen 0001, Lerrel Pinto, Li Fei-Fei 0001, Liam Tan, Linxi Fan, Lionel Ott, Lisa Lee, Luca Weihs, Magnum Chen, Marion Lepert, Marius Memmel, Masayoshi Tomizuka, Masha Itkina, Mateo Guaman Castro, Max Spero, Maximilian Du, Michael Ahn, Michael C. Yip, Mingtong Zhang 0003, Mingyu Ding, Minho Heo, Mohan Kumar Srirama, Mohit Sharma 0001, Moo Jin Kim, Naoaki Kanazawa, Nicklas Hansen 0001, Nicolas Heess, Nikhil J. Joshi, Niko Sünderhauf, Norman Di Palo, Nur Muhammad Shafiullah, Oier Mees, Oliver Kroemer, Osbert Bastani, Pannag R. Sanketi, Patrick Tree Miller, Patrick Yin, Paul Wohlhart, Peng Xu 0010, Peter David Fagan, Peter Mitrano, Pierre Sermanet, Pieter Abbeel, Priya Sundaresan, Qiuyu Chen, Rafael Rafailov, Ria Doshi, Roberto Martin Martin, Rohan Baijal, Rosario Scalise, Rose Hendrix, Roy Lin, Runjia Qian, Russell Mendonca, Rutav Shah, Ryan Hoque, Ryan Julian, Samuel Bustamante-Gomez, Sean Kirmani, Sergey Levine, Sherry Moore, Shikhar Bahl, Shivin Dass, Shubham D. Sonawani, Shuran Song, Sichun Xu, Siddhant Haldar, Siddharth Karamcheti, Simeon Adebola, Simon Guist, Soroush Nasiriany, Stefan Schaal, Stefan Welker, Stephen Tian, Subramanian Ramamoorthy, Sudeep Dasari, Suneel Belkhale, Sungjae Park, Suraj Nair 0003, Suvir Mirchandani, Takayuki Osa, Tanmay Gupta, Tatsuya Harada, Tatsuya Matsushima, Ted Xiao, Thomas Kollar, Tianhe Yu, Tianli Ding, Todor Davchev, Tony Z. Zhao, Travis Armstrong, Trevor Darrell, Trinity Chung, Vidhi Jain, Vincent Vanhoucke, Wolfram Burgard, Xiaolong Wang 0004, Xinghao Zhu, Xinyang Geng, Liangwei Xu, Yecheng Jason Ma 0001, Yejin Kim 0003, Yevgen Chebotar, Yilin Wu 0003, Yonatan Bisk, Yoonyoung Cho, Youngwoon Lee, Yuchen Cui, Yueh-Hua Wu, Yujin Tang, Yuke Zhu, Yunchu Zhang, Yunfan Jiang 0001, Yunshuang Li, Yunzhu Li, Yusuke Iwasawa, Yutaka Matsuo, Zehan Ma, Zichen Jeff Cui, Zichen Zhang 0016, Zipeng Lin	ICRA	66
2024	Conditionally Combining Robot Skills using Large Language Models abstract This paper combines two contributions. First, we introduce an extension of the Meta-World benchmark, which we call "Language-World," which allows a large language model to operate in a simulated robotic environment using semi-structured natural language queries and scripted skills described using natural language. By using the same set of tasks as Meta-World, Language-World results can be easily compared to Meta-World results, allowing for a point of comparison between recent methods using Large Language Models (LLMs) and those using Deep Reinforcement Learning. Second, we introduce a method we call Plan Conditioned Behavioral Cloning (PCBC), that allows finetuning the behavior of high-level plans using end-to-end demonstrations. Using Language-World, we show that PCBC is able to achieve strong performance in a variety of few-shot regimes, often achieving task generalization with as little as a single demonstration. We have made Language-World available as open-source software at https://github.com/krzentner/language-world/. K. R. Zentner, Ryan Julian, Brian Ichter, Gaurav S. Sukhatme	ICRA	4
2024	Inverse Submodular Maximization with Application to Human-in-the-Loop Multi-Robot Multi-Objective Coverage Control abstract We consider a new type of inverse combinatorial optimization, Inverse Submodular Maximization (ISM), for human-in-the-loop multi-robot coordination. Forward combinatorial optimization - solving a combinatorial problem given the reward (cost)-related parameters - is widely used in multi-robot coordination. In the standard pipeline, the reward (cost)-related parameters are designed offline by domain experts. These parameters are utilized for coordinating robots online. What if non-expert human supervisors desire to change these parameters during task execution to adapt to some new requirements? We are interested in the case where human supervisors can suggest what actions to take, and the robots need to change these internal parameters accordingly. We study such problems from the perspective of inverse combinatorial optimization, i.e., the process of finding parameters given solutions to the problem. Specifically, we propose a new formulation for ISM, in which we aim to find a new set of parameters that minimally deviate from the current parameters while causing a greedy algorithm to output actions which are the same as those desired by the human supervisors. We show that such problems can be formulated as a Mixed Integer Quadratic Program (MIQP) which is intractable for existing solvers when the problem size is large. We propose a new Branch & Bound algorithm to solve such problems. In numerical simulations, we demonstrate how to use ISM in multi-robot multi-objective coverage control, and we show that the proposed algorithm provides significant advantages in running time and peak memory usage compared to directly using an existing solver. Guangyao Shi, Gaurav S. Sukhatme	IROS	2
2024	Resilient Multi-Robot Multi-Target Tracking abstract We address the problem of ensuring resource availability in a networked multi-robot system performing distributed target tracking. Specifically, we consider a multi-target tracking scenario where the targets are driven by exogenous inputs that are unknown to the robots performing the tracking task. Robots track the positions of targets using a form of the Distributed Kalman Filter (DKF). We use the trace of each robot’s sensor measurement noise covariance matrix as a measure of its sensing quality. When a robot’s sensing quality deteriorates, the team’s communication graph is modified by adding edges such that the robot with deteriorating sensor quality may share information with other robots to improve the team’s target tracking ability. This computation is performed centrally and is designed to work without a large change in the number of active inter-robot communication links. Our method generates coordinates for the robots such the new communication graph can be realized in 3D. To achieve this, we propose two mixed integer semi-definite programming formulations, namely an ‘agent-centric’ strategy and a ‘team-centric’ strategy. We implement both formulations and a greedy, baseline strategy in simulation. Our simulation results show that the team-centric approach outperforms both agent-centric and greedy methods. Additionally, we show the effectiveness of our method in real-world settings through a multi-robot experiment performed in real-timeNote to Practitioners—This paper is motivated by the need to track multiple targets by means of a multi-robot team. When robots in the team experience degradation in their sensing quality our method reconfigures the team’s communication graph by repositioning robots. This allows the robot with deteriorated sensing quality to benefit from sensor measurements from its neighbors. In previous work, we solved this problem under the strong assumption the targets moved under inputs that were known to the tracking team. In this paper we remove this assumption. Our experiments show comparable tracking accuracy for both settings, suggesting that this strong assumption is not always necessary for good tracking results. Our method is straightforward to implement in Python using off-the-shelf optimization solvers. We assume that the team performs target tracking using a distributed algorithm, but has access to a powerful centralized base station to solve the computationally expensive underlying optimization problems. Developing a fully decentralized method and finding better ways to solve the underlying optimization problems are potential directions for future work. Ragesh K. Ramachandran, Nicole Fronda, James A. Preiss, Zhenghao Dai, Gaurav S. Sukhatme	IEEE Trans Autom. Sci. Eng.	5
2023	A Simple Approach for Visual Room Rearrangement: 3D Mapping and Semantic Search Brandon Trabucco, Gunnar A. Sigurdsson, Robinson Piramuthu, Gaurav S. Sukhatme, Ruslan Salakhutdinov	ICLR	4
2023	Fast and Scalable Signal Inference for Active Robotic Source Seeking abstract In active source seeking, a robot takes repeated measurements in order to locate a signal source in a cluttered and unknown environment. A key component of an active source seeking robot planner is a model that can produce estimates of the signal at unknown locations with uncertainty quantification. This model allows the robot to plan for future measurements in the environment. Traditionally, this model has been in the form of a Gaussian process, which has difficulty scaling and cannot represent obstacles. We propose a global and local factor graph model for active source seeking, which allows the model to scale to a large number of measurements and represent unknown obstacles in the environment. We combine this model with extensions to a highly scalable planner to form a system for large-scale active source seeking. We demonstrate that our approach outperforms baseline methods in both simulated and real robot experiments. Chris Denniston, Oriana Peltzer, Joshua Ott, Sung-Kyun Kim, Gaurav S. Sukhatme, Mykel J. Kochenderfer, Mac Schwager, Ali-akbar Agha-mohammadi	ICRA	6
2023	Efficiently Learning Small Policies for Locomotion and Manipulation abstract Neural control of memory-constrained, agile robots requires small, yet highly performant models. We leverage graph hyper networks to learn graph hyper policies trained with off-policy reinforcement learning resulting in networks that are two orders of magnitude smaller than commonly used networks yet encode policies comparable to those encoded by much larger networks trained on the same task. We show that our method can be appended to any off-policy reinforcement learning algorithm, without any change in hyperparameters, by showing results across locomotion and manipulation tasks. Further, we obtain an array of working policies, with differing numbers of parameters, allowing us to pick an optimal network for the memory constraints of a system. Training multiple policies with our method is as sample efficient as training a single policy. Finally, we provide a method to select the best architecture, given a constraint on the number of parameters. Project website: https://sites.google.com/usc.edu/graphhyperpolicy Shashank Hegde, Gaurav S. Sukhatme	ICRA	2
2023	Learned Parameter Selection for Robotic Information Gathering abstract When robots are deployed in the field for environmental monitoring they typically execute pre-programmed motions, such as lawnmower paths, instead of adaptive methods, such as informative path planning. One reason for this is that adaptive methods are dependent on parameter choices that are both critical to set correctly and difficult for the non-specialist to choose. Here, we show how to automatically configure a planner for informative path planning by training a reinforcement learning agent to select planner parameters at each iteration of informative path planning. We demonstrate our method with 37 instances of 3 distinct environments, and compare it against pure (end-to-end) reinforcement learning techniques, as well as approaches that do not use a learned model to change the planner parameters. Our method shows a 9.53% mean improvement in the cumulative reward across diverse environments when compared to end-to-end learning based methods; we also demonstrate via a field experiment how it can be readily used to facilitate high performance deployment of an information gathering robot. Chris Denniston, Gautam Salhotra, Akseli Kangaslahti, David A. Caron, Gaurav S. Sukhatme	IROS	5
2023	RREx-BoT: Remote Referring Expressions with a Bag of Tricks abstract Household robots operate in the same space for years. Such robots incrementally build dynamic maps that can be used for tasks requiring remote object localization. However, benchmarks in robot learning often test generalization through inference on tasks in unobserved environments. In an observed environment, locating an object is reduced to choosing from among all object proposals in the environment, which may number in the 100,000s. Armed with this intuition, using only a generic vision-language scoring model with minor modifications for 3d encoding and operating in an embodied environment, we demonstrate an absolute performance gain of 9.84% on remote object grounding above state of the art models for REVERIE and of 5.04% on FAO. When allowed to pre-explore an environment, we also exceed the previous state of the art pre-exploration method on REVERIE. Additionally, we demonstrate our model on a real-world TurtleBot platform, highlighting the simplicity and usefulness of the approach. Our analysis outlines a “bag of tricks” essential for accomplishing this task, from utilizing 3d coordinates and context, to gener-alizing vision-language models to large 3d search spaces. Gunnar A. Sigurdsson, Jesse Thomason, Gaurav S. Sukhatme, Robinson Piramuthu	IROS	3
2023	Alexa Arena: A User-Centric Interactive Platform for Embodied AI abstract We introduce Alexa Arena, a user-centric simulation platform to facilitate research in building assistive conversational embodied agents. Alexa Arena features multi-room layouts and an abundance of interactable objects. With user-friendly graphics and control mechanisms, the platform supports the development of gamified robotic tasks readily accessible to general human users, allowing high-efficiency data collection and EAI system evaluation. Along with the platform, we introduce a dialog-enabled task completion benchmark with online human evaluations. Qiaozi Gao, Govind Thattai, Suhaila M. Shakiah, Xiaofeng Gao 0002, Shreyas Pansare, Vasu Sharma, Gaurav S. Sukhatme, Hangjie Shi, Bofei Yang, Lucy Hu, Karthika Arumugam, Shui Hu, Matthew Wen, Dinakar Guthy, Shunan Chung, Rohan Khanna, Osman Ipek, Leslie Ball, Kate Bland, Heather Rocker, Michael Johnston, Reza Ghanadan, Dilek Hakkani-Tür, Premkumar Natarajan	NeurIPS	7
2023	Generating Behaviorally Diverse Policies with Latent Diffusion Models abstract Recent progress in Quality Diversity Reinforcement Learning (QD-RL) has enabled learning a collection of behaviorally diverse, high performing policies. However, these methods typically involve storing thousands of policies, which results in high space-complexity and poor scaling to additional behaviors. Condensing the archive into a single model while retaining the performance and coverage of the original collection of policies has proved challenging. In this work, we propose using diffusion models to distill the archive into a single generative model over policy parameters. We show that our method achieves a compression ratio of 13x while recovering 98% of the original rewards and 89% of the original humanoid archive coverage. Further, the conditioning mechanism of diffusion models allows for flexibly selecting and sequencing behaviors, including using language. Project website: https://sites.google.com/view/policydiffusion/home. Shashank Hegde, Sumeet Batra, K. R. Zentner, Gaurav S. Sukhatme	NeurIPS	4
2023	Synthesis of Large-Scale Instant IoT Networks abstract While most networks have long lifetimes, temporary network infrastructure is often useful for special events, pop-up retail, or disaster response. Aninstant IoTnetwork is one that is rapidly constructed, used for a few days, then dismantled. We consider the synthesis of instant IoT networks in urban settings. This synthesis problem must satisfy complex and competing constraints: sensor coverage, line-of-sight visibility, and network connectivity. The central challenge in our synthesis problem is quicklyscalingto large regions while producing cost-effective solutions. We explore two qualitatively different representations of the synthesis problems using satisfiability modulo convex optimization (SMC), and mixed-integer linear programming (MILP). The former is more expressive, for our problem, than the latter, but is less well-suited for solving optimization problems like ours. We show how to express our network synthesis in these frameworks. To scale to problem sizes beyond what these frameworks are capable of, we develop ahierarchical synthesistechnique that independently synthesizes networks in sub-regions of the deployment area, then combines these. We find that, while MILP outperforms SMC in some settings for smaller problem sizes, the fact that SMC's expressivity matches our problem ensures that it uniformly generates better quality solutions at larger problem sizes. Pradipta Ghosh, Jonathan Bunton, Dimitrios Pylorof, Marcos A. M. Vieira, Kevin S. Chan, Ramesh Govindan, Gaurav S. Sukhatme, Paulo Tabuada, Gunjan Verma	IEEE Trans. Mob. Comput.	7
2022	SSL Enables Learning from Sparse Rewards in Image-Goal Navigation Arjun Majumdar, Gunnar A. Sigurdsson, Robinson Piramuthu, Jesse Thomason, Dhruv Batra, Gaurav S. Sukhatme	ICML	6
2022	Probabilistic Inference of Simulation Parameters via Parallel Differentiable Simulation abstract Reproducing real world dynamics in simulation is critical for the development of new control and perception methods. This task typically involves the estimation of simu-lation parameter distributions from observed rollouts through an inverse inference problem characterized by multi-modality and skewed distributions. We address this challenging problem through a novel Bayesian inference approach that approximates a posterior distribution over simulation parameters given real sensor measurements. By extending the commonly used Gaus-sian likelihood model for trajectories via the multiple-shooting formulation, our gradient-based particle inference algorithm, Stein Variational Gradient Descent, is able to identify highly nonlinear, underactuated systems. We leverage GPU code gen-eration and differentiable simulation to evaluate the likelihood and its gradient for many particles in parallel. Our algorithm infers nonparametric distributions over simulation parame-ters more accurately than comparable baselines and handles constraints over parameters efficiently through gradient-based optimization. We evaluate estimation performance on several physical experiments. On an underactuated mechanism where a 7-DOF robot arm excites an object with an unknown mass configuration, we demonstrate how the inference technique can identify symmetries between the parameters and provide highly accurate predictions. Website: https://uscresl.github.io/prob-diff-sim Eric Heiden, Chris Denniston, David Millard 0001, Fabio Ramos 0001, Gaurav S. Sukhatme	ICRA	5
2022	Tracking Fast Trajectories with a Deformable Object using a Learned Model abstract We propose a method for robotic control of deformable objects using a learned nonlinear dynamics model. After collecting a dataset of trajectories from the real system, we train a recurrent neural network (RNN) to approximate its input-output behavior with a latent state-space model. The RNN internal state is low-dimensional enough to enable realtime nonlinear control methods. We demonstrate a closed-loop control scheme with the RNN model using a standard nonlinear state observer and model-predictive controller. We apply our method to track a highly dynamic trajectory with a point on the deformable object, in real time and on real hardware. Our experiments show that the RNN model captures the true system's frequency response and can be used to track trajectories outside the training distribution. In an ablation study, we find that the full method improves tracking accuracy compared to an open-loop version without the state observer. James A. Preiss, David Millard 0001, Gaurav S. Sukhatme	ICRA	4
2022	Inferring Articulated Rigid Body Dynamics from RGBD Video abstract Being able to reproduce physical phenomena ranging from light interaction to contact mechanics, simulators are becoming increasingly useful in more and more application domains where real-world interaction or labeled data are difficult to obtain. Despite recent progress, significant human effort is needed to configure simulators to accurately reproduce real-world behavior. We introduce a pipeline that combines inverse rendering with differentiable simulation to create digital twins of real-world articulated mechanisms from depth or RGB videos. Our approach automatically discovers joint types and estimates their kinematic parameters, while the dynamic properties of the overall mechanism are tuned to attain physically accurate simulations. Control policies optimized in our derived simulation transfer successfully back to the original system, as we demonstrate on a simulated system. Further, our approach accurately reconstructs the kinematic tree of an articulated mechanism being manipulated by a robot, and highly nonlinear dynamics of a real-world coupled pendulum mechanism. Website: https://eric-heiden.github.io/video2sim Eric Heiden, Ziang Liu 0002, Vibhav Vineet, Erwin Coumans, Gaurav S. Sukhatme	IROS	5
2022	Learning to Act with Affordance-Aware Multimodal Neural SLAM abstract Recent years have witnessed an emerging paradigm shift toward embodied artificial intelligence, in which an agent must learn to solve challenging tasks by interacting with its environment. There are several challenges in solving embodied multimodal tasks, including long-horizon planning, vision-and-language grounding, and efficient exploration. We focus on a critical bottleneck, namely the performance of planning and navigation. To tackle this challenge, we propose a Neural SLAM approach that, for the first time, utilizes several modalities for exploration, predicts an affordance-aware semantic map, and plans over it at the same time. This signif-icantly improves exploration efficiency, leads to robust long-horizon planning, and enables effective vision-and-language grounding. With the proposed Affordance-aware Multimodal Neural SLAM (AMSLAM) approach, we obtain more than 40% improvement over prior published work on the ALFRED benchmark and set a new state-of-the-art generalization per-formance at a success rate of 23.48% on the test unseen scenes. Zhiwei Jia, Kaixiang Lin, Qiaozi Gao, Govind Thattai, Gaurav S. Sukhatme	IROS	6
2022	Asynchronous Real-time Decentralized Multi-Robot Trajectory Planning abstract We present a novel overconstraining and constraint-discarding method for asynchronous, real-time, decentralized, multi-robot trajectory planning that ensures collision avoidance. Our approach utilizes communication between robots. The communication medium is best-effort: messages may be dropped, re-ordered or delayed. Robots conservatively constrain themselves against others assuming they may be working with outdated information, and discard constraints when they receive update messages from others. Our method can augment existing synchronized decentralized receding horizon planning algorithms that utilize separating hyperplanes for collision avoidance thereby making them applicable to asynchronous setups. As an example, we extend an existing model predictive control based, synchronized, decentralized multi-robot planner using our method. We show our method's effectiveness under asynchronous planning and imperfect communication by comparing our extension to the base version. Our extension does not result in any collisions or synchronization-induced deadlocks to which the base version is prone. Baskin Senbaslar, Gaurav S. Sukhatme	IROS	2
2022	Efficient Multi-Task Learning via Iterated Single-Task Transfer abstract In order to be effective general purpose machines in real world environments, robots not only will need to adapt their existing manipulation skills to new circumstances, they will need to acquire entirely new skills on-the-fly. One approach to achieving this capability is via Multi-task Reinforcement Learning (MTRL). Most recent work in MTRL trains a single policy to solve all tasks at once. In this work, we investigate the feasibility of instead training separate policies for each task, and only transferring from a task once the policy for it has finished training. We describe a method of finding near optimal sequences of transfers to perform in this setting, and use it to show that performing the optimal sequence of transfer is competitive with other MTRL methods on the Meta World MT10 benchmark. Lastly, we describe a method for finding nearly optimal transfer sequences during training that is able to improve on training each task from scratch. K. R. Zentner, Ujjwal Puri, Yulun Zhang 0002, Ryan Julian, Gaurav S. Sukhatme	IROS	5
2022	Parameter Estimation for Deformable Objects in Robotic Manipulation Tasks David Millard 0001, James A. Preiss, Jernej Barbic, Gaurav S. Sukhatme	ISRR	4
2022	The Role of Heterogeneity in Autonomous Perimeter Defense Problems Aviv Adler, Oscar Mickelin, Ragesh K. Ramachandran, Gaurav S. Sukhatme, Sertac Karaman	WAFR	4
2022	Sensing the Sensor: Estimating Camera Properties with Minimal Information abstract Public outdoor surveillance cameras often have limited metadata describing their properties. Frequently, a public camera’s precise position, orientation, focal length, and image center are unknown; these attributes are necessary to precisely pinpoint the location of events seen in the camera. In this article, we ask: what is the minimal information needed to accurately estimate these properties for public cameras? We show, using a judicious combination of projective geometry, neural networks, and crowd-sourced annotations from human workers, that it is possible to, for example, localize 95% of the cameras in our test data set to within 12 m using a single image taken from the camera. This performance is an order of magnitude better than PoseNet, a state-of-the-art neural network that needs significantly more information than our approach, and can only estimate position and orientation (and not other properties). Finally, we show that the camera’s inferred pose and properties can help design a number of virtual sensors , all of which have good accuracy. Pradipta Ghosh, Hang Qiu 0001, Marcos A. M. Vieira, Gaurav S. Sukhatme, Ramesh Govindan	ACM Trans. Sens. Networks	5
2022	Introduction to the Special Section on Resilience in Networked Robotic Systems abstract The 17 papers in this special section focus on resilience in networked robotic systems. This collection of articles aims to provide a deeper understanding of resilience as it pertains to multirobot systems, and to disseminate the current advances in designing and operating networked robotic systems. We understand resilience to be a characteristic that enables amultirobot system to withstand or overcome unexpected adverse conditions or shocks, and unknown, unmodeled disturbances. It refers to the contingent nature of the robots’ behaviors that is aimed at preserving their functionality or minimizing the time periods during which their functionality is compromised. The papers explore new algorithmic and mathematical foundations toward resilience. Amanda Prorok, Vijay Kumar 0001, Brian M. Sadler, Gaurav S. Sukhatme	IEEE Trans. Robotics	4
2022	Resilient Monitoring in Heterogeneous Multi-Robot Systems Through Network Reconfiguration abstract We propose a framework for resilience in a networked heterogeneous multirobot team subject to resource failures. Each robot in the team is equipped with resources that it shares with its neighbors, which are identified based on the team’s communication graph. Additionally, each robot in the team executes a task, whose performance depends on the resources to which it has access. When a resource on a particular robot becomes unavailable ( e.g., a camera ceases to function), the team optimally reconfigures its communication network so that the robots affected by the failure can continue their tasks. We focus on a monitoring task, where robots individually estimate the state of an exogenous process. We encode the end-to-end effect of a robot’s resource loss on the monitoring performance of the team by defining a new stronger notion of observability—one-hop observability. By abstracting the impact that low-level individual resources have on the task performance through the notion of one-hop observability, our framework leads to the principled reconfiguration of information flow in the team to effectively replace the lost resource on one robot with information from another, as long as certain conditions are met. Network reconfiguration is converted to the problem of selecting edges to be modified in the system’s communication graph after a resource failure has occurred. A controller based on finite-time convergence control barrier functions drives each robot to a spatial location that enables the communication links of the modified graph. We validate the effectiveness of our framework by deploying it on a team of differential-drive robots estimating the position of a group of quadrotors. Ragesh K. Ramachandran, Pietro Pierpaoli, Magnus Egerstedt, Gaurav S. Sukhatme	IEEE Trans. Robotics	4
2021	NeuralSim: Augmenting Differentiable Simulators with Neural Networks abstract Differentiable simulators provide an avenue for closing the sim-to-real gap by enabling the use of efficient, gradient-based optimization algorithms to find the simulation parameters that best fit the observed sensor readings. Nonetheless, these analytical models can only predict the dynamical behavior of systems for which they have been designed. In this work, we study the augmentation of a novel differentiable rigid-body physics engine via neural networks that is able to learn nonlinear relationships between dynamic quantities and can thus model effects not accounted for in traditional simulators. Such augmentations require less data to train and generalize better compared to entirely data-driven models. Through extensive experiments, we demonstrate the ability of our hybrid simulator to learn complex dynamics involving frictional contacts from real data, as well as match known models of viscous friction, and present an approach for automatically discovering useful augmentations. We show that, besides benefiting dynamics modeling, inserting neural networks can accelerate model-based control architectures. We observe a ten-fold speedup when replacing the QP solver inside a model-predictive gait controller for quadruped robots with a neural network, allowing us to significantly improve control delays as we demonstrate in real-hardware experiments. We publish code, additional results and videos from our experiments on our project webpage at https://sites.google.com/usc.edu/neuralsim. Eric Heiden, David Millard 0001, Erwin Coumans, Yizhou Sheng, Gaurav S. Sukhatme	ICRA	5
2021	Adaptive Sampling using POMDPs with Domain-Specific Considerations abstract We investigate improving Monte Carlo Tree Search based solvers for Partially Observable Markov Decision Processes (POMDPs), when applied to adaptive sampling problems. We propose improvements in rollout allocation, the action exploration algorithm, and plan commitment. The first allocates a different number of rollouts depending on how many actions the agent has taken in an episode. We find that rollouts are more valuable after some initial information is gained about the environment. Thus, a linear increase in the number of rollouts, i.e. allocating a fixed number at each step, is not appropriate for adaptive sampling tasks. The second alters which actions the agent chooses to explore when building the planning tree. We find that by using knowledge of the number of rollouts allocated, the agent can more effectively choose actions to explore. The third improvement is in determining how many actions the agent should take from one plan. Typically, an agent will plan to take the first action from the planning tree and then call the planner again from the new state. Using statistical techniques, we show that it is possible to greatly reduce the number of rollouts by increasing the number of actions taken from a single planning tree without affecting the agent’s final reward. Finally, we demonstrate experimentally, on simulated and real aquatic data from an underwater robot, that these improvements can be combined, leading to better adaptive sampling. The code for this work is available at https://github.com/uscresl/AdaptiveSamplingPOMCP. Gautam Salhotra, Chris Denniston, David A. Caron, Gaurav S. Sukhatme	ICRA	4
2020	Rapid Top-Down Synthesis of Large-Scale IoT Networks abstract Advances in optimization and constraint satisfaction techniques, together with the availability of elastic computing resources, have spurred interest in large-scale network verification and synthesis. Motivated by this, we consider the top-down synthesis of ad-hoc IoT networks for disaster response and search and rescue operations. This synthesis problem must satisfy complex and competing constraints: sensor coverage, line-of-sight visibility, and network connectivity. The central challenge in our synthesis problem is quickly scaling to large regions while producing cost-effective solutions. We explore a representation of the synthesis problems using a novel constraint satisfaction paradigm, satisfiability modulo convex optimization (SMC). We choose SMC because it matches the expressivity needs for our network synthesis. To scale to large problem sizes, we develop a hierarchical synthesis technique that independently synthesizes networks in sub-regions of the deployment area, then combines these. Our experiments show that SMC consistently generates better quality solutions than a baseline synthesis approach based on Mixed Integer Linear Programming (MILP). Pradipta Ghosh, Jonathan Bunton, Dimitrios Pylorof, Marcos A. M. Vieira, Kevin S. Chan, Ramesh Govindan, Gaurav S. Sukhatme, Paulo Tabuada, Gunjan Verma	ICCCN	7
2020	Sample Factory: Egocentric 3D Control from Pixels at 100000 FPS with Asynchronous Reinforcement Learning abstract Increasing the scale of reinforcement learning experiments has allowed researchers to achieve unprecedented results in both training sophisticated agents for video games, and in sim-to-real transfer for robotics. Typically such experiments rely on large distributed systems and require expensive hardware setups, limiting wider access to this exciting area of research. In this work we aim to solve this problem by optimizing the efficiency and resource utilization of reinforcement learning algorithms instead of relying on distributed computation. We present the "Sample Factory", a high-throughput training system optimized for a single-machine setting. Our architecture combines a highly efficient, asynchronous, GPU-based sampler with off-policy correction techniques, allowing us to achieve throughput higher than $10^5$ environment frames/second on non-trivial control problems in 3D without sacrificing sample efficiency. We extend Sample Factory to support self-play and population-based training and apply these techniques to train highly capable agents for a multiplayer first-person shooter game. Github: https://github.com/alex-petrenko/sample-factory Aleksei Petrenko, Zhehui Huang, Tushar Kumar, Gaurav S. Sukhatme, Vladlen Koltun	ICML	4
2020	Meta Learning via Learned Loss abstract Typically, loss functions, regularization mechanisms and other important aspects of training parametric models are chosen heuristically from a limited set of options. In this paper, we take the first step towards automating this process, with the view of producing models which train faster and more robustly. Concretely, we present a meta-learning method for learning parametric loss functions that can generalize across different tasks and model architectures. We develop a pipeline for “meta-training” such loss functions, targeted at maximizing the performance of the model trained under them. The loss landscape produced by our learned losses significantly improves upon the original task-specific losses in both supervised and reinforcement learning tasks. Furthermore, we show that our meta-learning framework is flexible enough to incorporate additional information at meta-train time. This information shapes the learned loss function such that the environment does not need to provide this information during meta-test time. We make our code available at https://sites.google.com/view/mlthree Sarah Bechtle, Artem Molchanov, Yevgen Chebotar, Edward Grefenstette, Ludovic Righetti, Gaurav S. Sukhatme, Franziska Meier	ICPR	6
2020	Physics-based Simulation of Continuous-Wave LIDAR for Localization, Calibration and Tracking abstract Light Detection and Ranging (LIDAR) sensors play an important role in the perception stack of autonomous robots, supplying mapping and localization pipelines with depth measurements of the environment. While their accuracy outperforms other types of depth sensors, such as stereo or time-of-flight cameras, the accurate modeling of LIDAR sensors requires laborious manual calibration that typically does not take into account the interaction of laser light with different surface types, incidence angles and other phenomena that significantly influence measurements. In this work, we introduce a physically plausible model of a 2D continuous-wave LIDAR that accounts for the surface-light interactions and simulates the measurement process in the Hokuyo URG-04LX LIDAR. Through automatic differentiation, we employ gradient-based optimization to estimate model parameters from real sensor measurements. Eric Heiden, Ziang Liu 0002, Ragesh K. Ramachandran, Gaurav S. Sukhatme	ICRA	4
2020	Resilience in multi-robot target tracking through reconfiguration abstract We address the problem of maintaining resource availability in a networked multi-robot system performing distributed target tracking. In our model, robots are equipped with sensing and computational resources enabling them to track a target's position using a Distributed Kalman Filter (DKF). We use the trace of each robot's sensor measurement noise covariance matrix as a measure of sensing quality. When a robot's sensing quality deteriorates, the systems communication graph is modified by adding edges such that the robot with deteriorating sensor quality may share information with other robots to improve the team's target tracking ability. This computation is performed centrally and is designed to work without a large change in the number of active communication links. We propose two mixed integer semi-definite programming formulations (an `agent-centric' strategy and a `team-centric' strategy) to achieve this goal. We implement both formulations and a greedy strategy in simulation and show that the team-centric strategy outperforms the agent-centric and greedy strategies. Ragesh K. Ramachandran, Nicole Fronda, Gaurav S. Sukhatme	ICRA	3
2020	Persistent Connected Power Constrained Surveillance with Unmanned Aerial Vehicles abstract Persistent surveillance with aerial vehicles (drones) subject to connectivity and power constraints is a relatively uncharted domain of research. To reduce the complexity of multi-drone motion planning, most state-of-the-art solutions ignore network connectivity and assume unlimited battery power. Motivated by this and advances in optimization and constraint satisfaction techniques, we introduce a new persistent surveillance motion planning problem for multiple drones that incorporates connectivity and power consumption constraints. We use a recently developed constrained optimization tool (Satisfiability Modulo Convex Optimization (SMC)) that has the expressivity needed for this problem. We show how to express the new persistent surveillance problem in the SMC framework. Our analysis of the formulation based on a set of simulation experiments illustrates that we can generate the desired motion planning solution within a couple of minutes for small teams of drones (up to 5) confined to a 7 × 7 × 1 grid-space. Pradipta Ghosh, Paulo Tabuada, Ramesh Govindan, Gaurav S. Sukhatme	IROS	4
2020	Resilient Coverage: Exploring the Local-to-Global Trade-off abstract We propose a centralized control framework to select suitable robots from a heterogeneous pool and place them at appropriate locations to monitor a region for events of interest. In the event of a robot failure, our framework repositions robots in a user-defined local neighborhood of the failed robot to compensate for the coverage loss. If repositioning robots locally fails to attain a user-specified level of desired coverage, the central controller augments the team with additional robots from the pool. The size of the local neighborhood around the failed robot and the desired coverage over the region are two objectives that can be varied to achieve a user-specified balance. We investigate the trade-off between the coverage compensation achieved through local repositioning and the computation required to plan the new robot locations. We also study the relationship between the size of the local neighborhood and the number of additional robots added to the team for a given user-specified level of desired coverage. Through extensive simulations and an experiment with a team of seven quadrotors we verify the effectiveness of our framework. We show that to reach a high level of coverage in a neighborhood with a large robot population, it is more efficient to enlarge the neighborhood size, instead of adding additional robots and repositioning them. Ragesh K. Ramachandran, Lifeng Zhou 0001, James A. Preiss, Gaurav S. Sukhatme	IROS	4
2020	Pac-Man is Overkill abstract Pursuit-Evasion Game (PEG) consists of a team of pursuers trying to capture one or more evaders. PEG is important due to its application in surveillance, search and rescue, disaster robotics, boundary defense and so on. In general, PEG requires exponential time to compute the minimum number of pursuers to capture an evader. To mitigate this, we have designed a parallel optimal algorithm to minimize the capture time in PEG. Given a discrete topology, this algorithm also outputs the minimum number of pursuers to capture an evader. A classic example of PEG is the popular arcade game, Pac-Man. Although Pac-Man topology has almost 300 nodes, our algorithm can handle this. We show that Pac-Man is overkill, i.e., given the Pac-Man game topology, Pac-Man game contains more pursuers/ghosts (four) than it is necessary (two) to capture evader/Pac-man. We evaluate the proposed algorithm on many different topologies. Renato Fernando dos Santos, Ragesh K. Ramachandran, Marcos A. M. Vieira, Gaurav S. Sukhatme	IROS	4
2020	Mobile Robot Localization under Non-Gaussian noise using Correntropy Similarity Metric abstract In this paper, we study the localization problem under non-Gaussian noise. In particular, we consider systems that can be represented by a state transition and a measurement component. The state transition indicates how the system evolves given a control variable. The measurement component compares, for a given state, the received and predicted measurements. Here we consider a radio based range sensor which is the primary source of non-Gaussian noise in the system. We solve the problem using a MHE (Maximum Horizon Estimator) with a correntropy similarity metric. Given a time window, the MHE seeks the best set of states that explains the system for the received measurements. Moreover, the main advantage of a MHE is that it allows the re-estimation of past states. Additionally, the correntropy is a similarity metric that, given the amount of error in the estimation, behaves as L2, L1 or L0 norms and has been successfully used in many applications under non-Gaussian noise. We evaluate our proposed method using both simulated and real data. The results show that correntropy is able to work well in comparison with other methods in presence of impulsive noise. Elerson Rubens da Silva Santos, Marcos A. M. Vieira, Gaurav S. Sukhatme	IROS	3
2019	Coordinating multi-robot systems through environment partitioning for adaptive informative sampling abstract As robotic platforms have become more capable and autonomous, they have increasingly been utilized in time sensitive applications such as search and rescue. To that end, we have developed a system for teams of robots to efficiently explore an environment while taking sensor measurements. The system utilizes an information seeking algorithm that generates high priority points of interest based on the highest expected information gained per distance travelled. In order to coordinate multiple robots, the system partitions the area into different regions according to the effort needed to explore each region. Robots are assigned different regions to measure in order to minimize repetition of work and reduce interference between each robot.We present an information rate adaptive sampling approach for tasking robots within an environment to gather sensor measurements. We evaluated our approach within a simulation environment with one to four robots. Multiple robots are coordinated through our region segmentation approach. The data shows efficiency gains through the use of adaptive information gain rate tasking above a naïve closest point approach. We also see positive results from using the region segmentation technique. We further the experimentation by testing the algorithm on real world robots and verify the results in real world experimentation. Nicholas Fung, John G. Rogers III, Carlos Nieto, Henrik I. Christensen, Stephanie Kemna, Gaurav S. Sukhatme	ICRA	6
2019	Estimating Metric Scale Visual Odometry from Videos using 3D Convolutional Networks abstract We present an end-to-end deep learning approach for performing metric scale-sensitive regression tasks such visual odometry with a single camera and no additional sensors. We propose a novel 3D convolutional architecture, 3DC-VO, that can leverage temporal relationships over a short moving window of images to estimate linear and angular velocities. The network makes local predictions on stacks of images that can be integrated to form a full trajectory. We apply 3DC-VO to the KITTI visual odometry benchmark and the task of estimating a pilot’s control inputs from a first-person video of a quadrotor flight. Our method exhibits increased accuracy relative to comparable learning-based algorithms trained on monocular images. We also show promising results for quadrotor control input prediction when trained on a new dataset collected with a UAV simulator. Alexander S. Koumis, James A. Preiss, Gaurav S. Sukhatme	IROS	3
2019	Sim-to-(Multi)-Real: Transfer of Low-Level Robust Control Policies to Multiple Quadrotors abstract Quadrotor stabilizing controllers often require careful, model-specific tuning for safe operation. We use reinforcement learning to train policies in simulation that transfer remarkably well to multiple different physical quadrotors. Our policies are low-level, i.e., we map the rotorcrafts' state directly to the motor outputs. The trained control policies are very robust to external disturbances and can withstand harsh initial conditions such as throws. We show how different training methodologies (change of the cost function, modeling of noise, use of domain randomization) might affect flight performance. To the best of our knowledge, this is the first work that demonstrates that a simple neural network can learn a robust stabilizing low-level quadrotor controller (without the use of a stabilizing PD controller) that is shown to generalize to multiple quadrotors. The video of our experiments can be found at https://sites.google.com/view/sim-to-multi-quad. Artem Molchanov, Wolfgang Hönig, James A. Preiss, Nora Ayanian, Gaurav S. Sukhatme	IROS	6
2019	Resilience by Reconfiguration: Exploiting Heterogeneity in Robot Teams abstract We propose a method to maintain high resource availability in a networked heterogeneous multi-robot system subject to resource failures. In our model, resources such as sensing and computation are available on robots. The robots are engaged in a joint task using these pooled resources. When a resource on a particular robot becomes unavailable (e.g., a sensor ceases to function), the system automatically reconfigures so that the robot continues to have access to this resource by communicating with other robots. Specifically, we consider the problem of selecting edges to be modified in the system's communication graph after a resource failure has occurred. We define a metric that allows us to characterize the quality of the resource distribution in the network represented by the communication graph. Upon a resource becoming unavailable due to failure, we reconFigure the network so that the resource distribution is brought as close to the maximal resource distribution as possible without a large change in the number of active inter-robot communication links. Our approach uses mixed integer semi-definite programming to achieve this goal. We employ a simulated annealing method to compute a spatial formation that satisfies the inter-robot distances imposed by the topology, along with other constraints. Our method can compute a communication topology, spatial formation, and formation change motion planning in a few seconds. We validate our method in simulation and real-robot experiments with a team of seven quadrotors. Ragesh K. Ramachandran, James A. Preiss, Gaurav S. Sukhatme	IROS	3
2018	Will Distributed Computing Revolutionize Peace? The Emergence of Battlefield IoT abstract An upcoming frontier for distributed computing might literally save lives in future military operations. In civilian scenarios, significant efficiencies were gained from interconnecting devices into networked services and applications that automate much of everyday life from smart homes to intelligent transportation. The ecosystem of such applications and services is collectively called the Internet of Things (IoT). Can similar benefits be gained in a military context by developing an IoT for the battlefield? This paper describes unique challenges in such a context as well as potential risks, mitigation strategies, and benefits. Tarek F. Abdelzaher, Nora Ayanian, Tamer Basar, Suhas N. Diggavi, Jana Diesner, Deepak Ganesan, Ramesh Govindan, Susmit Jha, Tancrède Lepoint, Benjamin M. Marlin, Klara Nahrstedt, David M. Nicol, Ragunathan Rajkumar, Stephen Russell 0001, Sanjit A. Seshia, Fei Sha, Prashant J. Shenoy, Mani Srivastava 0001, Gaurav S. Sukhatme, Ananthram Swami, Paulo Tabuada, Don Towsley, Nitin H. Vaidya, Venugopal V. Veeravalli	ICDCS	19
2018	Gradient-Informed Path Smoothing for Wheeled Mobile Robots abstract Planning smooth trajectories is important for the safe, efficient and comfortable operation of mobile robots, such as wheeled robots moving in crowded environments or cars moving at high speed. Asymptotically optimal sampling-based motion planners can be used to generate such trajectories. However, to achieve the necessary efficiency for the realtime operation of robots, one often uses their initial feasible trajectories or the trajectories of non-optimal motion planners instead, typically after a post-smoothing step. We propose a gradient-informed post-smoothing algorithm, called GRIPS, that deforms given trajectories by locally optimizing the placement of vertices while satisfying the system's kinodynamic constraints. We show experimentally that GRIPS typically produces trajectories of significantly smaller length and higher smoothness than several existing post-smoothing algorithms. Eric Heiden, Luigi Palmieri, Sven Koenig, Kai Oliver Arras, Gaurav S. Sukhatme	ICRA	5
2018	Pilot Surveys for Adaptive Informative Sampling abstract Adaptive sampling has been shown to be an effective method for modeling environmental fields, such as algae concentrations in the ocean. In adaptive sampling, a robot adapts its sampling trajectory based on data that it is collecting. This data is often aggregated into models, using techniques such as Gaussian Process (G P) regression. The (hyper-)parameters for these models need to be manually set or, ideally, estimated from data. For GP regression, hyperparameters are typically estimated using prior data. This paper addresses the case where initial hyperparameters need to be estimated, but no prior data is available. Without prior data or accurately pre-defined hyperparameters, adaptive sampling techniques may fail, because there is no good model to base path planning decisions on. One method of gathering data is to perform a pilot survey. This survey needs to select informative samples for initiating the model, but without having a model to determine where best to sample. In this work, we evaluate four pilot surveys, which use a softmax function on the distance between waypoints and previously sampled data for waypoint selection. Simulation results show that pilot surveys that maximize waypoint spread over randomization lead to more stable estimation of GP hyperparameters, and create accurate models more quickly. Stephanie Kemna, Oliver Kroemer, Gaurav S. Sukhatme	ICRA	3
2018	Learning Manipulation Graphs from Demonstrations Using Multimodal Sensory Signals abstract Complex contact manipulation tasks can be decomposed into sequences of motor primitives. Individual primitives often end with a distinct contact state, such as inserting a screwdriver tip into a screw head or loosening it through twisting. To achieve robust execution, the robot should be able to verify that the primitive's goal has been reached as well as disambiguate it from erroneous contact states. In this paper, we introduce and evaluate a framework to autonomously construct manipulation graphs from manipulation demonstrations. Our manipulation graphs include sequences of motor primitives for performing a manipulation task as well as corresponding contact state information. The sensory models for the contact states allow the robot to verify the goal of each motor primitive as well as detect erroneous contact changes. The proposed framework was experimentally evaluated on grasping, unscrewing, and insertion tasks on a Barrett arm and hand equipped with two BioTacs. The results of our experiments indicate that the learned manipulation graphs achieve more robust manipulation executions by confirming sensory goals as well as discovering and detecting novel failure modes. Oliver Kroemer, Gerald E. Loeb, Gaurav S. Sukhatme, Stefan Schaal	ICRA	4
2018	Solving Markov Decision Processes with Reachability Characterization from Mean First Passage Times abstract A new mechanism for efficiently solving the Markov decision processes (MDPs) is proposed in this paper. We introduce the notion of reachability landscape where we use the Mean First Passage Time (MFPT) as a means to characterize the reachability of every state in the state space. We show that such reachability characterization very well assesses the importance of states and thus provides a natural basis for effectively prioritizing states and approximating policies. Built on such a novel observation, we design two new algorithms - Mean First Passage Time based Value Iteration (MFPT-VI) and Mean First Passage Time based Policy Iteration (MFPT-PI) - that have been modified from the state-of-the-art solution methods. To validate our design, we have performed numerical evaluations in robotic decision-making scenarios, by comparing the proposed new methods with corresponding classic baseline mechanisms. The evaluation results showed that MFPT-VI and MFPT-PI have outperformed the state-of-the-art solutions in terms of both practical runtime and number of iterations. Aside from the advantage of fast convergence, this new solution method is intuitively easy to understand and practically simple to implement. Shoubhik Debnath, Lantao Liu, Gaurav S. Sukhatme	IROS	3
2018	Accelerating Goal-Directed Reinforcement Learning by Model Characterization abstract We propose a hybrid approach aimed at improving the sample efficiency in goal-directed reinforcement learning. We do this via a two-step mechanism where firstly, we approximate a model from Model-Free reinforcement learning. Then, we leverage this approximate model along with a notion of reachability using Mean First Passage Times to perform Model-Based reinforcement learning. Built on such a novel observation, we design two new algorithms - Mean First Passage Time based Q-Learning (MFPT-Q)and Mean First Passage Time based DYNA (MFPT-DYNA), that have been fundamentally modified from the state-of-the-art reinforcement learning techniques. Preliminary results have shown that our hybrid approaches converge with much fewer iterations than their corresponding state-of-the-art counterparts and therefore requiring much fewer samples and much fewer training trials to converge. Shoubhik Debnath, Gaurav S. Sukhatme, Lantao Liu	IROS	2
2018	Trajectory Planning for Quadrotor Swarms abstract We describe a method for multirobot trajectory planning in known, obstacle-rich environments. We demonstrate our approach on a quadrotor swarm navigating in a warehouse setting. Our method consists of following three stages: 1) roadmap generation that generates sparse roadmaps annotated with possible interrobot collisions; 2) discrete planning that finds valid execution schedules in discrete time and space; 3) continuous refinement that creates smooth trajectories. We account for the downwash effect of quadrotors, allowing safe flight in dense formations. We demonstrate computational efficiency in simulation with up to 200 robots and physical plausibility with an experiment on 32 nano-quadrotors. Our approach can compute safe and smooth trajectories for hundreds of quadrotors in dense environments with obstacles in a few minutes. Wolfgang Hönig, James A. Preiss, T. K. Satish Kumar, Gaurav S. Sukhatme, Nora Ayanian	IEEE Trans. Robotics	4
2017	Combining Model-Based and Model-Free Updates for Trajectory-Centric Reinforcement Learning abstract Reinforcement learning algorithms for real-world robotic applications must be able to handle complex, unknown dynamical systems while maintaining data-efficient learning. These requirements are handled well by model-free and model-based RL approaches, respectively. In this work, we aim to combine the advantages of these approaches. By focusing on time-varying linear-Gaussian policies, we enable a model-based algorithm based on the linear-quadratic regulator that can be integrated into the model-free framework of path integral policy improvement. We can further combine our method with guided policy search to train arbitrary parameterized policies such as deep neural networks. Our simulation and real-world experiments demonstrate that this method can solve challenging manipulation tasks with comparable or better performance than model-free methods while maintaining the sample efficiency of model-based methods. Yevgen Chebotar, Karol Hausman, Marvin Zhang, Gaurav S. Sukhatme, Stefan Schaal, Sergey Levine	ICML	4
2017	Multi-robot coordination through dynamic Voronoi partitioning for informative adaptive sampling in communication-constrained environments abstract Autonomous underwater vehicles (AUVs) are cost- and time-efficient systems for environmental sampling. Informative adaptive sampling has been shown to be an effective method of sampling a lake or ocean for environmental modeling. In this paper, we focus on multi-robot coordination for informative adaptive sampling. We use a dynamic Voronoi partitioning approach whereby the vehicles, in a decentralized fashion, repeatedly calculate weighted Voronoi partitions for the space. Each vehicle then runs informative adaptive sampling within their partition. The vehicles can request surfacing events to share data between vehicles. Simulation results show that the addition of the coordination with dynamic Voronoi partitioning results in obtaining higher quality models faster. Thus we created a decentralized, multi-robot coordination approach for informative, adaptive sampling of unknown environments. Stephanie Kemna, John G. Rogers III, Carlos Nieto-Granda, Stuart Young, Gaurav S. Sukhatme	ICRA	5
2017	Feature selection for learning versatile manipulation skills based on observed and desired trajectories abstract For a manipulation skill to be applicable to a wide range of scenarios, it must generalize between different objects and object configurations. Robots should therefore learn skills that adapt to features describing the objects being manipulated. Most of these object features will however be irrelevant for generalizing the skill and, hence, the robot should select a small set of relevant features for adapting the skill. We use a framework for learning versatile manipulation skills that adapt to a sparse set of object features. Skills are initially learned from demonstrations and subsequently improved using reinforcement learning. The robot also learns a meta prior over the features' relevances to guide the feature selection process. In this paper, we explore using either desired trajectories or observed trajectories for selecting the relevant features. The framework was evaluated on placing, tilting, and wiping tasks. The evaluations showed that using the desired trajectories to select the relevant features lead to better skill learning performance. Oliver Kroemer, Gaurav S. Sukhatme	ICRA	2
2017	Informative planning and online learning with sparse Gaussian processes abstract A big challenge in environmental monitoring is the spatiotemporal variation of the phenomena to be observed. To enable persistent sensing and estimation in such a setting, it is beneficial to have a time-varying underlying environmental model. Here we present a planning and learning method that enables an autonomous marine vehicle to perform persistent ocean monitoring tasks by learning and refining an environmental model. To alleviate the computational bottleneck caused by large-scale data accumulated, we propose a framework that iterates between a planning component aimed at collecting the most information-rich data, and a sparse Gaussian Process learning component where the environmental model and hyperparameters are learned online by taking advantage of only a subset of data that provides the greatest contribution. Our simulations with ground-truth ocean data shows that the proposed method is both accurate and efficient. Kai-Chieh Ma, Lantao Liu, Gaurav S. Sukhatme	ICRA	3
2017	Crazyswarm: A large nano-quadcopter swarm abstract We define a system architecture for a large swarm of miniature quadcopters flying in dense formation indoors. The large number of small vehicles motivates novel design choices for state estimation and communication. For state estimation, we develop a method to reliably track many small rigid bodies with identical motion-capture marker arrangements. Our communication infrastructure uses compressed one-way data flow and supports a large number of vehicles per radio. We achieve reliable flight with accurate tracking (<; 2 cm mean position error) by implementing the majority of computation onboard, including sensor fusion, control, and some trajectory planning. We provide various examples and empirically determine latency and tracking performance for swarms with up to 49 vehicles. James A. Preiss, Wolfgang Hönig, Gaurav S. Sukhatme, Nora Ayanian	ICRA	3
2017	Planning high-speed safe trajectories in confidence-rich maps abstract Planning safe, high-speed trajectories in unknown environments remains a major roadblock on the way toward achieving fast autonomous flight. Current state-of-the-art planning approaches use sampling-based methods or trajectory optimization to obtain fast trajectories, whose safety is evaluated by taking into account the current state estimate of the environment. In unknown environments, however, this leads to numerous stops caused by the need for re-planning the trajectory due to unexpected obstacles. In this paper, we propose to use an active perception paradigm for planning. We predict the future uncertainty of the map and optimize trajectories to minimize re-planning risk. This leads to faster and safer trajectories. We evaluate the proposed planning approach in a series of simulation experiments, which show that we are able to achieve safer trajectories with a smaller number of re-planning stops and faster speeds. Eric Heiden, Karol Hausman, Gaurav S. Sukhatme, Ali-akbar Agha-mohammadi	IROS	3
2017	A spatio-temporal representation for the orienteering problem with time-varying profits abstract We consider an orienteering problem (OP) where an agent needs to visit a series (possibly a subset) of depots, from which the maximal accumulated profits are desired within given limited time budget. Different from most existing works where the profits are assumed to be static, in this work we investigate a variant that has arbitrary time-dependent profits. Specifically, the profits to be collected change over time and they follow different (e.g., independent) time-varying functions. The problem is of inherent nonlinearity and difficult to solve by existing methods. To tackle the challenge, we present a simple and effective framework that incorporates time-variations into the fundamental planning process. Specifically, we propose a deterministic spatio-temporal representation where both spatial description and temporal logic are unified into one routing topology. By employing existing basic sorting and searching algorithms, the routing solutions can be computed in an extremely efficient way. The proposed method is easy to implement and extensive numerical results show that our approach is time efficient and generates near-optimal solutions. Zhibei Ma, Lantao Liu, Gaurav S. Sukhatme	IROS	4
2017	Downwash-aware trajectory planning for large quadrotor teams abstract We describe a method for formation-change trajectory planning for large quadrotor teams in obstacle-rich environments. Our method decomposes the planning problem into two stages: a discrete planner operating on a graph representation of the workspace, and a continuous refinement that converts the non-smooth graph plan into a set of Ck-continuous trajectories, locally optimizing an integral-squared-derivative cost. We account for the downwash effect, allowing safe flight in dense formations. We demonstrate the computational efficiency in simulation with up to 200 robots and the physical plausibility with an experiment with 32 nano-quadrotors. Our approach can compute safe and smooth trajectories for hundreds of quadrotors in dense environments with obstacles in a few minutes. James A. Preiss, Wolfgang Hönig, Nora Ayanian, Gaurav S. Sukhatme	IROS	4
2017	Confidence-Rich Grid Mapping Ali-akbar Agha-mohammadi, Eric Heiden, Karol Hausman, Gaurav S. Sukhatme	ISRR	4
2017	Reachability and Differential Based Heuristics for Solving Markov Decision Processes Shoubhik Debnath, Lantao Liu, Gaurav S. Sukhatme	ISRR	3
2017	Multi-Modal Imitation Learning from Unstructured Demonstrations using Generative Adversarial Nets abstract Imitation learning has traditionally been applied to learn a single task from demonstrations thereof. The requirement of structured and isolated demonstrations limits the scalability of imitation learning approaches as they are difficult to apply to real-world scenarios, where robots have to be able to execute a multitude of tasks. In this paper, we propose a multi-modal imitation learning framework that is able to segment and imitate skills from unlabelled and unstructured demonstrations by learning skill segmentation and imitation learning jointly. The extensive simulation results indicate that our method can efficiently separate the demonstrations into individual skills and learn to imitate them using a single multi-modal policy. Karol Hausman, Yevgen Chebotar, Stefan Schaal, Gaurav S. Sukhatme, Joseph J. Lim	NIPS	4
2017	Interactive Perception: Leveraging Action in Perception and Perception in Action abstract Recent approaches in robot perception follow the insight that perception is facilitated by interaction with the environment. These approaches are subsumed under the term Interactive Perception (IP). This view of perception provides the following benefits. First, interaction with the environment creates a rich sensory signal that would otherwise not be present. Second, knowledge of the regularity in the combined space of sensory data and action parameters facilitates the prediction and interpretation of the sensory signal. In this survey, we postulate this as a principle for robot perception and collect evidence in its support by analyzing and categorizing existing work in this area. We also provide an overview of the most important applications of IP. We close this survey by discussing remaining open questions. With this survey, we hope to help define the field of Interactive Perception and to provide a valuable resource for future research. Jeannette Bohg, Karol Hausman, Bharath Sankaran, Oliver Brock, Danica Kragic, Stefan Schaal, Gaurav S. Sukhatme	IEEE Trans. Robotics	7
2017	Generalized Topology Control for Nonholonomic Teams With Discontinuous Interactions abstract In this paper, we consider the problem of general topology control in multirobot systems with nonholonomic kinematics. Our contribution is twofold: We first demonstrate the correctness of topology control under the assumption that the network topology can switch arbitrarily and that potential-based mobility is discontinuous with respect to topology changes; we then demonstrate that a multirobot team under the above listed conditions continues to achieve topology control when actuator saturation is applied and in the presence of arbitrary discontinuous (and possibly nonpairwise) exogenous objectives. Simulation results are given to corroborate our theoretical findings. Ryan K. Williams, Andrea Gasparri, Giovanni Ulivi, Gaurav S. Sukhatme	IEEE Trans. Robotics	4
2016	Self-calibrating multi-sensor fusion with probabilistic measurement validation for seamless sensor switching on a UAV abstract Fusing data from multiple sensors on-board a mobile platform can significantly augment its state estimation abilities and enable autonomous traversals of different domains by adapting to changing signal availabilities. However, due to the need for accurate calibration and initialization of the sensor ensemble as well as coping with erroneous measurements that are acquired at different rates with various delays, multi-sensor fusion still remains a challenge. In this paper, we introduce a novel multi-sensor fusion approach for agile aerial vehicles that allows for measurement validation and seamless switching between sensors based on statistical signal quality analysis. Moreover, it is capable of self-initialization of its extrinsic sensor states. These initialized states are maintained in the framework such that the system can continuously self-calibrate. We implement this framework on-board a small aerial vehicle and demonstrate the effectiveness of the above capabilities on real data. As an example, we fuse GPS data, ultra-wideband (UWB) range measurements, visual pose estimates, and IMU data. Our experiments demonstrate that our system is able to seamlessly filter and switch between different sensors modalities during run time. Karol Hausman, Stephan Weiss 0002, Roland Brockers, Larry H. Matthies, Gaurav S. Sukhatme	ICRA	5
2016	Self-supervised regrasping using spatio-temporal tactile features and reinforcement learning abstract We introduce a framework for learning regrasping behaviors based on tactile data. First, we present a grasp stability predictor that uses spatio-temporal tactile features collected from the early-object-lifting phase to predict the grasp outcome with a high accuracy. Next, the trained predictor is used to supervise and provide feedback to a reinforcement learning algorithm that learns the required grasp adjustments based on tactile feedback. Our results gathered over more than 50 hours of real robot experiments indicate that the robot is able to predict the grasp outcome with 93% accuracy. In addition, the robot is able to improve the grasp success rate from 42% when randomly grasping an object to up to 97% when allowed to regrasp the object in case of a predicted failure. Yevgen Chebotar, Karol Hausman, Gaurav S. Sukhatme, Stefan Schaal	IROS	4
2016	Occlusion-aware multi-robot 3D tracking abstract We introduce an optimization-based control approach that enables a team of robots to cooperatively track a target using onboard sensing. In this setting, the robots are required to estimate their own positions as well as concurrently track the target. Our probabilistic method generates controls that minimize the expected uncertainty of the target. Additionally, our method efficiently reasons about occlusions between robots and takes them into account for the control generation. We evaluate our approach in a number of experiments in which we simulate a team of quadrotor robots flying in three-dimensional space to track a moving target on the ground. We compare our method to other state-of-the-art approaches represented by the random sampling technique, lattice planning method, and our previous method. Our experimental results indicate that our method achieves up to 8 times smaller maximum tracking error and up to 2 times smaller average tracking error than the next best approach in the presented scenarios. Karol Hausman, Gregory Kahn, Sachin Patil, Jörg Müller 0004, Kenneth Y. Goldberg, Pieter Abbeel, Gaurav S. Sukhatme	IROS	7
2016	An information-driven and disturbance-aware planning method for long-term ocean monitoring abstract We propose an efficient path planning method for an autonomous underwater vehicle (AUV) used for the long-range and long-term ocean monitoring. We consider both the spatio-temporal variations of ocean phenomena and the disturbances caused by ocean currents, and design an approach integrating the information-theoretic and decision-theoretic planning frameworks. Specifically, the information-theoretic component employs a hierarchical structure and plans the most informative observation way-points for reducing the uncertainty of ocean phenomena modeling and prediction; whereas the decision-theoretic component plans local motions by taking into account the non-stationary ocean current disturbances. We validated the method through simulations with real ocean data. Kai-Chieh Ma, Lantao Liu, Gaurav S. Sukhatme	IROS	3
2016	Contact localization on grasped objects using tactile sensing abstract Manipulation tasks often require robots to make contact between a grasped tool and another object in the robot's environment. The ability to detect and estimate the positions and directions of these contact points is crucial for monitoring the progress of the task, and detecting failures. In this paper, we present a data-driven approach for detecting and localizing contacts between a grasped object and the environment using tactile sensing. We explore framing the contact localization as both a regression and a classification problem and train neural networks accordingly to estimate the contact parameters. We also compare the neural networks with Gaussian process regression and support vector machine classification with spatio-temporal hierarchical matching pursuit feature learning. We evaluate the presented approach using hundreds of contact events on eighteen objects with different shapes, sizes and material properties. The experiments show that the neural network approach can learn to localize contact events for individual objects with a mean absolute error of less than 2.5 cm for the positions and less than 10° for the directions. Artem Molchanov, Oliver Kroemer, Gaurav S. Sukhatme	IROS	4
2016	Online trajectory optimization to improve object recognition abstract We present an online trajectory optimization approach that optimizes a trajectory such that object recognition performance is improved. Inspired by prior work, we formulate the optimization as a derivative-free stochastic optimization, allowing us to express the cost function in an arbitrary way. The cost function is defined such that information acquisition of target objects is improved, while simultaneously moving towards the goal point. We show the evaluation of our approach on a quadrotor platform in simulation as well as on a real robot. The results show that by using an online optimization approach recognition accuracy is greatly improved, but more importantly the optimized trajectory reduces the uncertainty of the posterior class distribution greatly. Hence, verifying that the optimized trajectory collects more valuable information. Christian Potthast, Gaurav S. Sukhatme	IROS	2
2016	Designing Sparse Reliable Pose-Graph SLAM: A Graph-Theoretic Approach Kasra Khosoussi, Gaurav S. Sukhatme, Shoudong Huang, Gamini Dissanayake	WAFR	2
2015	Active articulation model estimation through interactive perception abstract We introduce a particle filter-based approach to representing and actively reducing uncertainty over articulated motion models. The presented method provides a probabilistic model that integrates visual observations with feedback from manipulation actions to best characterize a distribution of possible articulation models. We evaluate several action selection methods to efficiently reduce the uncertainty about the articulation model. The full system is experimentally evaluated using a PR2 mobile manipulator. Our experiments demonstrate that the proposed system allows for intelligent reasoning about sparse, noisy data in a number of common manipulation scenarios. Karol Hausman, Scott Niekum, Sarah Osentoski, Gaurav S. Sukhatme	ICRA	4
2015	Active drifters: Towards a practical multi-robot system for ocean monitoring abstract We propose a method for controlling multiple active drifters in the presence of external forcing induced by the ocean. Our active drifters have one actuator: they can lower and raise their drogues in depth. By exploiting the vertically stratified nature of ocean currents, we show how classical multi-robot tasks (spreading out and aggregation) can be accomplished by the multi-drifter system. Tests with a realistic simulation based on an ocean model suggest that a practical implementation of active drifters which aggregate and disperse in the coastal ocean could be realized through our control method with relatively inexpensive components. Specifically, we are able to show that over a 90 day deployment a significant fraction of drifters can be made to aggregate in few clusters suitable for recovery. Artem Molchanov, Andreas Breitenmoser, Gaurav S. Sukhatme	ICRA	3
2015	Multi-step planning for robotic manipulation abstract Most current systems capable of robotic object manipulation involve ad hoc assumptions about the order of operations necessary to achieve a task, and usually have no mechanism to predict how earlier decisions will affect the quality of the solution later. Solving this problem is sometimes referred to as combined task and motion planning. We propose that multi-step planning, a technique previously applied in some other domains, is an effective way to address the question of combined task and motion planning. We demonstrate the technique on a complex motion planning problem involving a two-armed robot (PR2) and an articulated object (folding chair) where our planner naturally discovers extra steps that are necessary to satisfy kinematic constraints of the problem. We also propose some further extensions to our algorithm that we believe will make it an extremely powerful technique in this domain. Max Pflueger, Gaurav S. Sukhatme	ICRA	2
2015	Global connectivity control for spatially interacting multi-robot systems with unicycle kinematics abstract In this paper, we consider the problem of connectivity maintenance in multi-robot systems with unicycle kinematics. While previous work has approached this problem through local control techniques, we propose a solution which achieves global connectivity maintenance under nonholonomic constraints. In addition, our formulation only requires intermittent estimation of algebraic connectivity, and accommodates discontinuous spatial interactions among robots. Specifically, we extend a decision-based link maintenance framework to unicycle kinematics and discontinuous potential-based interaction, by exploiting techniques from nonsmooth analysis. Then, we couple this extension with an existing connectivity estimation technique which yields an estimate with tunable precision in finite time, achieving our result. To illustrate the correctness of our methods, we provide a brief simulation result that closes the paper. Ryan K. Williams, Andrea Gasparri, Gaurav S. Sukhatme, Giovanni Ulivi	ICRA	3
2015	Observability in topology-constrained multi-robot target tracking abstract In this paper, we consider the problem of controlling a multi-robot team in order to track a mobile target with unknown dynamics. Our contribution in this context is a non-linear observability analysis of the target tracking network, which yields insights into the topological and actuation conditions necessary for the underlying state estimation problem. We demonstrate a metric of observability which allows us to determine robot inputs that maximize observability to improve team localization. Combining the observability metric with topological control then yields a robust target tracking solution. We close the paper with a simulation example which demonstrates the ability of our solution to cope with scenarios in which relative measurements poorly distinguish the target. Ryan K. Williams, Gaurav S. Sukhatme	ICRA	2
2015	Interactive affordance map building for a robotic task abstract We describe a technique to build an affordance map interactively for robotic tasks. Affordances are predicted by a trained classifier using geometric features extracted from objects. Based on 2D occupancy grid, a Markov Random Field (MRF) model builds an affordance map with relational affordance with neighboring cells. The quality of the affordance map is refined by sequences of interactive manipulations selected from the model to yield the highest reduction in uncertainty. David Inkyu Kim, Gaurav S. Sukhatme	IROS	2
2015	Active Multi-view Object Recognition and Online Feature Selection Christian Potthast, Andreas Breitenmoser, Fei Sha, Gaurav S. Sukhatme	ISRR (2)	4
2015	CARLOC: Precise Positioning of Automobiles abstract Precise positioning of an automobile to within lane-level precision can enable better navigation and context-awareness. However, GPS by itself cannot provide such precision in obstructed urban environments. In this paper, we present a system called CARLOC for lane-level positioning of automobiles. CARLOC uses three key ideas in concert to improve positioning accuracy: it uses digital maps to match the vehicle to known road segments; it uses vehicular sensors to obtain odometry and bearing information; and it uses crowd-sourced location of estimates of roadway landmarks that can be detected by sensors available in modern vehicles. CARLOC unifies these ideas in a probabilistic position estimation framework, widely used in robotics, called the sequential Monte Carlo method. Through extensive experiments on a real vehicle, we show that CARLOC achieves sub-meter positioning accuracy in an obstructed urban setting, an order-of-magnitude improvement over a high-end GPS device. Yurong Jiang, Hang Qiu 0001, Matthew McCartney, Gaurav S. Sukhatme, Marco Gruteser, Fan Bai 0002, Donald Grimm, Ramesh Govindan	SenSys	4
2015	Poster: CARLOC: Precisely Tracking Automobile Position abstract Precise positioning of an automobile to within lane-level precision can enable better navigation and context-awareness. However, GPS by itself cannot provide such precision in obstructed urban environments. In this paper, we present a system called CARLOC for lane-level positioning of automobiles. CARLOC uses three key ideas in concert to improve positioning accuracy: it uses digital maps to match the vehicle to known road segments; it uses vehicular sensors to obtain odometry and bearing information; and it uses crowd-sourced location of estimates of roadway landmarks that can be detected by sensors available in modern vehicles. CARLOC unifies these ideas in a probabilistic position estimation framework, widely used in robotics, called the sequential Monte Carlo method. Through extensive experiments on a real vehicle, we show that CARLOC achieves sub-meter positioning accuracy in an obstructed urban setting, an order-of-magnitude improvement over a high-end GPS device. Yurong Jiang, Hang Qiu 0001, Matthew McCartney, Gaurav S. Sukhatme, Marco Gruteser, Fan Bai 0002, Donald Grimm, Ramesh Govindan	SenSys	4
2015	Using Manipulation Primitives for Object Sorting in Cluttered Environments abstract This paper explores the idea of manipulation-aided perception and grasping in the context of sorting small objects on a cluttered tabletop. We present a robust pipeline that combines perception and manipulation to accurately sort objects by some property (e.g., color, size, shape, etc.). The pipeline uses two motion primitives to manipulate the scene in ways that help the robot to improve its perception and grasps. This results in the ability to sort cluttered object piles accurately. We also present an implementation on the PR2 robot that applies our algorithm to sort Duplo bricks by color and size, and compare our method to brick sorting without the aid of manipulation. The experimental results demonstrate the benefits of our approach, particularly in environments with a high degree of clutter. Jörg Müller 0004, Gaurav S. Sukhatme	IEEE Trans Autom. Sci. Eng.	3
2015	Rigidity-Preserving Team Partitions in Multiagent Networks abstract Motivated by the strong influence network rigidity has on collaborative systems, in this paper, we consider the problem of partitioning a multiagent network into two sub-teams, a bipartition, such that the resulting sub-teams are topologically rigid. In this direction, we determine the existence conditions for rigidity-preserving bipartitions, and provide an iterative algorithm that identifies such partitions in polynomial time. In particular, the relationship between rigid graph partitions and the previously identified Z-link edge structure is given, yielding a feasible direction for graph search. Adapting a supergraph search mechanism, we then detail a methodology for discerning graphs cuts that represent valid rigid bipartitions. Next, we extend our methods to a decentralized context by exploiting leader election and an improved graph search to evaluate feasible cuts using only local agent-to-agent communication. Finally, full algorithm details and pseudocode are provided, together with simulation results that verify correctness and demonstrate complexity. Daniela Carboni, Ryan K. Williams, Andrea Gasparri, Giovanni Ulivi, Gaurav S. Sukhatme	IEEE Trans. Cybern.	5
2015	Decentralized and Parallel Constructionsfor Optimally Rigid Graphs in $\mathbb{R}^2$ abstract In this paper, we address the decentralized and parallel construction of rigid graphs in the plane that optimize an edge-weighted objective function under cardinality constraints. Two auction-based algorithms to solve this problem in a decentralized fashion are first proposed. Centered around the notion of leader election, the first approach finds an optimal solution through a greedy bidding, while the second approach provides a sub-optimal solution which reduces complexity according to a sliding mode parameter. Then, by exploiting certain local structural properties of graph rigidity, a parallelization to build a portion of the optimal solution in constant time is derived. A theoretical characterization of algorithm performance is provided together with complexity analysis. Finally, simulation results are presented to corroborate the theoretical findings. Andrea Gasparri, Ryan K. Williams, Attilio Priolo, Gaurav S. Sukhatme	IEEE Trans. Mob. Comput.	4
2015	Distributed Data Fusion for Multirobot Search abstract This paper presents novel data fusion methods that enable teams of vehicles to perform target search tasks without guaranteed communication. Techniques are introduced for merging estimates of a target's position from vehicles that regain contact after long periods of time, and a fully distributed team-planning algorithm is proposed, which utilizes limited shared information as it becomes available. The proposed data fusion techniques are shown to avoid overcounting information, which ensures that combining data from different vehicles will not decrease the performance of the search. Motivated by the underwater search domain, a realistic underwater acoustic communication channel is used to determine the probability of successful data transfer between two locations. The channel model is integrated into a simulation of multiple autonomous vehicles in both open water and harbor environments. The results demonstrate that the proposed distributed coordination techniques provide performance competitive with full communication. Geoffrey A. Hollinger, Srinivas Yerramalli, Sanjiv Singh, Urbashi Mitra, Gaurav S. Sukhatme	IEEE Trans. Robotics	5
2014	Trajectory learning for human-robot scientific data collection abstract We propose an integrated learning and planning framework that leverages knowledge from a human user along with prior information about the environment to generate trajectories for scientific data collection. The proposed framework combines principles from probabilistic planning with uncertainty modeling through nonparametric Bayesian methods to refine trajectories for execution by autonomous vehicles. The resulting techniques allow for trajectories specified by a user to be modified for reduced risk of collision and increased reliability. We test our approach in the underwater ocean monitoring domain, and we show that the proposed framework reduces the risk of collision with ship traffic by as much as 51% for an autonomous underwater vehicle operating in ocean currents. This work provides insight into the tools necessary for combining human-robot interaction with autonomous navigation. Geoffrey A. Hollinger, Gaurav S. Sukhatme	ICRA	2
2014	Semantic labeling of 3D point clouds with object affordance for robot manipulation abstract When a robot is deployed it needs to understand the nature of its surroundings. In this paper, we address the problem of semantic labeling 3D point clouds by object affordance (e.g., `pushable', `liftable'). We propose a technique to extract geometric features from point cloud segments and build a classifier to predict associated object affordances. With the classifier, we have developed an algorithm to enhance object segmentation and reduce manipulation uncertainty by iterative clustering, along with minimizing labeling entropy. Our incremental multiple view merging technique shows improved object segmentation. The novel feature of our approach is the semantic labeling that can be directly applied to manipulation planning. In our experiments with 6 affordance labels, an average of 81.8% accuracy of affordance prediction is achieved. We demonstrate refined object segmentation by applying the classifier to data from the PR2 robot using a Microsoft Kinect in an indoor office environment. David Inkyu Kim, Gaurav S. Sukhatme	ICRA	2
2014	Decentralized algorithms for optimally rigid network constructions abstract In this paper, we address the construction of optimally rigid networks that minimize an edge-weighted objective function over a planar graph. We propose two auction-based algorithms to solve this problem in a fully decentralized way. The first approach finds an optimal solution at the cost of high communication complexity; the second approach provides a sub-optimal solution while reducing the computational burden according to a sliding mode parameter ζ, yielding a tradeoff between complexity and optimality. A theoretical characterization of the optimality of the first algorithm is provided, and a closed form for the maximum gap between the optimal solution and the sub-optimal solution is also given. Simulation results are presented to corroborate the theoretical findings. Attilio Priolo, Ryan K. Williams, Andrea Gasparri, Gaurav S. Sukhatme	ICRA	4
2014	Risk-aware trajectory generation with application to safe quadrotor landing abstract In navigation tasks, mobile robots often have to deal with substantial uncertainty due to imperfect actuators and noisy sensor measurements. In this paper, we consider the problem of online trajectory generation for safe navigation in the presence of state uncertainty and the resulting deviations from the desired trajectory. Our approach combines probabilistic estimation of the a priori collision risk with efficient trajectory generation, exploiting the differential flatness of many robotic systems in an explicitly constrained polynomial trajectory representation. Through trajectory optimization, our approach allows to flexibly trade off risk against, for example, the duration of the trajectory. It is computationally efficient because each optimization step has polynomial complexity. In contrast to other approaches, our method can also optimize the trajectory duration and supports cost functions that facilitate higher-order smoothness of the trajectory. Our experiments demonstrate the performance of the approach and show that our trajectories result in substantially lower collisions probabilities compared to minimum-snap trajectories in a quadrotor landing task. Jörg Müller 0004, Gaurav S. Sukhatme	IROS	2
2014	A probabilistic framework for next best view estimation in a cluttered environment Christian Potthast, Gaurav S. Sukhatme	J. Vis. Commun. Image Represent.	2
2014	Hierarchical Approaches to Estimate Energy Expenditure Using Phone-Based Accelerometers abstract Physical inactivity is linked with increase in risk of cancer, heart disease, stroke, and diabetes. Walking is an easily available activity to reduce sedentary time. Objective methods to accurately assess energy expenditure from walking that is normalized to an individual would allow tailored interventions. Current techniques rely on normalization by weight scaling or fitting a polynomial function of weight and speed. Using the example of steady-state treadmill walking, we present a set of algorithms that extend previous work to include an arbitrary number of anthropometric descriptors. We specifically focus on predicting energy expenditure using movement measured by mobile phone-based accelerometers. The models tested include nearest neighbor models, weight-scaled models, a set of hierarchical linear models, multivariate models, and speed-based approaches. These are compared for prediction accuracy as measured by normalized average root mean-squared error across all participants. Nearest neighbor models showed highest errors. Feature combinations corresponding to sedentary energy expenditure, sedentary heart rate, and sex alone resulted in errors that were higher than speed-based models and nearest-neighbor models. Size-based features such as BMI, weight, and height produced lower errors. Hierarchical models performed better than multivariate models when size-based features were used. We used the hierarchical linear model to determine the best individual feature to describe a person. Weight was the best individual descriptor followed by height. We also test models for their ability to predict energy expenditure with limited training data. Hierarchical models outperformed personal models when a low amount of training data were available. Speed-based models showed poor interpolation capability, whereas hierarchical models showed uniform interpolation capabilities across speeds. Harshvardhan Vathsangam, E. Todd Schroeder, Gaurav S. Sukhatme	IEEE J. Biomed. Health Informatics	3
2014	Determining the Time Delay Between Inertial and Visual Sensor Measurements abstract We examine the problem of determining the relative time delay between IMU and camera data streams. The primary difficulty is that the correspondences between measurements from the sensors are not initially known, and hence, the time delay cannot be computed directly. We instead formulate time delay calibration as a registration problem, and introduce a calibration algorithm that operates by aligning curves in a three-dimensional orientation space. Results from simulation studies and from experiments with real hardware demonstrate that the delay can be accurately calibrated. Jonathan Kelly, Nicholas Roy, Gaurav S. Sukhatme	IEEE Trans. Robotics	3
2014	Evaluating Network Rigidity in Realistic Systems: Decentralization, Asynchronicity, and Parallelization abstract In this paper, we consider the problem of evaluating the rigidity of a planar network, while satisfying common objectives of real-world systems: decentralization, asynchronicity, and parallelization. The implications that rigidity has in fundamental multirobot problems, e.g., guaranteed formation stability and relative localizability, motivates this study. We propose the decentralization of the pebble game algorithm of Jacobs et al. , which is an O(n2) method that determines the generic rigidity of a planar network. Our decentralization is based on asynchronous messaging and distributed memory, coupled with auctions for electing leaders to arbitrate rigidity evaluation. Further, we provide a parallelization that takes inspiration from gossip algorithms to yield significantly reduced execution time and messaging. An analysis of the correctness, finite termination, and complexity is given, along with a simulated application in decentralized rigidity control. Finally, we provide Monte Carlo analysis in a Contiki networking environment, illustrating the real-world applicability of our methods, and yielding a bridge between rigidity theory and realistic interacting systems. Ryan K. Williams, Andrea Gasparri, Attilio Priolo, Gaurav S. Sukhatme	IEEE Trans. Robotics	4
2013	Hierarchical probabilistic regression for AUV-based adaptive sampling of marine phenomena abstract Marine phenomena such as algal blooms can be detected using in situ measurements onboard autonomous underwater vehicles (AUVs), but understanding plankton ecology and community structure requires retrieval and analysis of water specimens. This process requires shipboard or manual sample collection, followed by onshore lab analysis which is time-consuming. Better understanding of the relationship between the observable environmental features and organism abundance would allow more precisely targeted sampling and thereby save time. In this work, we present an approach to learn and improve models that predict this relationship. Coupled with recent advances in AUV technology allowing selective retrieval of water samples, this constitutes a new paradigm in biological sampling. We use organism abundance models along with spatial models of environmental features learned immediately after AUV deployments to compute spatial distributions of organisms in the coastal ocean purely from in situ AUV data. We use Gaussian process regression along with the unscented transform to fuse the two models, obtaining both the mean and variance of the organism abundance estimates. The uncertainty in organism abundance predictions is used in a sampling strategy to selectively acquire new water specimens that improves the organism abundance models. Simulation results are presented demonstrating the advantage of performing hierarchical probabilistic regression. After the validation through simulation, we show predictions of organism abundance from models learned on lab-analyzed water sample data, and AUV survey data. Jnaneshwar Das, Julio B. J. Harvey, Frédéric Py, Harshvardhan Vathsangam, Rishi Graham, Kanna Rajan, Gaurav S. Sukhatme	ICRA	7
2013	Learning uncertainty models for reliable operation of Autonomous Underwater Vehicles abstract We discuss the problem of learning uncertainty models of ocean processes to assist in the operation of Autonomous Underwater Vehicles (AUVs) in the ocean. We focus on the prediction of ocean currents, which have significant effect on the navigation of AUVs. Available models provide accurate prediction of ocean currents, but they typically do not provide confidence estimates of these predictions. We propose augmenting existing prediction methods with variance measures based on Gaussian Process (GP) regression. We show that commonly used measures of variance in GPs do not accurately reflect errors in ocean current prediction, and we propose an alternative uncertainty measure based on interpolation variance. We integrate these measures of uncertainty into a probabilistic planner running on an AUV during a field deployment in the Southern California Bight. Our experiments demonstrate that the proposed uncertainty measures improve the safety and reliability of AUVs operating in the coastal ocean. Geoffrey A. Hollinger, Arvind Pereira, Gaurav S. Sukhatme	ICRA	3
2013	Learning task error models for manipulation abstract Precise kinematic forward models are important for robots to successfully perform dexterous grasping and manipulation tasks, especially when visual servoing is rendered infeasible due to occlusions. A lot of research has been conducted to estimate geometric and non-geometric parameters of kinematic chains to minimize reconstruction errors. However, kinematic chains can include non-linearities, e.g. due to cable stretch and motor-side encoders, that result in significantly different errors for different parts of the state space. Previous work either does not consider such non-linearities or proposes to estimate non-geometric parameters of carefully engineered models that are robot specific. We propose a data-driven approach that learns task error models that account for such unmodeled non-linearities. We argue that in the context of grasping and manipulation, it is sufficient to achieve high accuracy in the task relevant state space. We identify this relevant state space using previously executed joint configurations and learn error corrections for those. Therefore, our system is developed to generate subsequent executions that are similar to previous ones. The experiments show that our method successfully captures the non-linearities in the head kinematic chain (due to a counterbalancing spring) and the arm kinematic chains (due to cable stretch) of the considered experimental platform, see Fig. 1. The feasibility of the presented error learning approach has also been evaluated in independent DARPA ARM-S testing contributing to successfully complete 67 out of 72 grasping and manipulation tasks. Peter Pastor, Mrinal Kalakrishnan, Jonathan Binney, Jonathan Kelly, Ludovic Righetti, Gaurav S. Sukhatme, Stefan Schaal	ICRA	6
2013	An investigation on the accuracy of Regional Ocean Models through field trials abstract Recent efforts in mission planning for underwater vehicles have utilised predictive models to aid in navigation, optimal path planning and drive opportunistic sampling. Although these models provide information at a unprecedented resolutions and have proven to increase accuracy and effectiveness in multiple campaigns, most are deterministic in nature. Thus, predictions cannot be incorporated into probabilistic planning frameworks, nor do they provide any metric on the variance or confidence of the output variables. In this paper, we provide an initial investigation into determining the confidence of ocean model predictions based on the results of multiple field deployments of two autonomous underwater vehicles. For multiple missions of two autonomous gliders conducted over a two-month period in 2011, we compare actual vehicle executions to simulations of the same missions through the Regional Ocean Modeling System in an ocean region off the coast of southern California. This comparison provides a qualitative analysis of the current velocity predictions for areas within the selected deployment region. Ultimately, we present a spatial heat-map of the correlation between the ocean model predictions and the actual mission executions. Knowing where the model provides unreliable predictions can be incorporated into planners to increase the utility and application of the deterministic estimations. Ryan N. Smith, Jonathan Kelly, Kimia Nazarzadeh, Gaurav S. Sukhatme	ICRA	4
2013	Locally constrained connectivity control in mobile robot networks abstract In this paper, we consider the problem of controlling the connectivity of a network of mobile agents under local topology constraints and proximity-limited communication. The inverse iteration algorithm for spectral analysis is formulated in a distributed manner to allow each agent to estimate a component of global network connectivity, improving on the convergence rate issues of previous approaches. Potential-based controls drive the agents to maximize connectivity under local degree constraints, maintain established links to guarantee connectivity, and avoid collisions. To achieve constraint satisfaction we exploit a switched model of interaction that regulates link addition through symmetric, repulsive potentials between constraint violators, enforcing discernment in communication through spatial organization. Simulations of connectivity estimation as well as agent aggregation and leader-following applications demonstrate the ability of our proposed methods to generate connectivity maximizing, constraint-aware self organization. Ryan K. Williams, Gaurav S. Sukhatme	ICRA	2
2013	Topology-constrained flocking in locally interacting mobile networks abstract In this paper, we consider the problem of controlling a network of locally interacting mobile agents, subject to a set of non-local topology constraints, towards a group flocking objective. As opposed to switching network links directly in the space of discrete graphs, yielding a divide in spatial configuration and communication topology, we regulate topology through mobility, enabling adjacent agents to retain or deny links spatially on the basis of constraint satisfaction. Specifically, we propose a distributed formulation consisting of a switching control and smooth potential fields for local link discrimination and flocking, coupled with consensus-based coordination over proposed topology changes, yielding transitions in communication that respect non-local constraints and correspond to agent configuration. An analysis of the interplay between the topology consensus and constraint composition, together with a Lyapunov-like convergence argument guarantees the flocking, collision avoidance, and constraint satisfaction properties of the system. Finally, simulations of a novel constrained coordination scenario highlight the correctness and applicability of our proposed methods. Ryan K. Williams, Gaurav S. Sukhatme	ICRA	2
2013	Interactive environment exploration in clutter abstract Robotic environment exploration in cluttered environments is a challenging problem. The number and variety of objects present not only make perception very difficult but also introduce many constraints for robot navigation and manipulation. In this paper, we investigate the idea of exploring a small, bounded environment (e.g., the shelf of a home refrigerator) by prehensile and non-prehensile manipulation of the objects it contains. The presence of multiple objects results in partial and occluded views of the scene. This inherent uncertainty in the scene's state forces the robot to adopt an observe-plan-act strategy and interleave planning with execution. Objects occupying the space and potentially occluding other hidden objects are rearranged to reveal more of the unseen area. The environment is considered explored when the state (free or occupied) of every voxel in the volume is known. The presented algorithm can be easily adapted to real world problems like object search, taking inventory, and mapping. We evaluate our planner in simulation using various metrics like planning time, number of actions required, and length of planning horizon. We then present an implementation on the PR2 robot and use it for object search in clutter. Thomas Rühr, Michael Beetz, Gaurav S. Sukhatme	IROS	4
2013	Decentralized generic rigidity evaluation in interconnected systems abstract In this paper, we consider the problem of evaluating the generic rigidity of an interconnected system in the plane, without a priori knowledge of the network's topological properties. We propose the decentralization of the pebble game algorithm of Jacobs et. al., an O(n2) method that determines the generic rigidity of a planar network. Our decentralization is based on asynchronous inter-agent message-passing and a distributed memory architecture, coupled with consensus-based auctions for electing leaders in the system. We provide analysis of the asynchronous messaging structure and its interaction with leader election, and Monte Carlo simulations demonstrating complexity and correctness. Finally, a novel rigidity evaluation and control scenario in the accompanying media illustrates the applicability of our proposed algorithm. Ryan K. Williams, Andrea Gasparri, Attilio Priolo, Gaurav S. Sukhatme	IROS	4
2013	An autonomous Wireless Networked Robotics System for backbone deployment in highly-obstructed environments Marcos A. M. Vieira, Ramesh Govindan, Gaurav S. Sukhatme	Ad Hoc Networks	3
2013	Mitigating multi-path fading in a mobile mesh network Marcos A. M. Vieira, Matthew E. Taylor, Prateek Tandon 0002, Ramesh Govindan, Gaurav S. Sukhatme, Milind Tambe	Ad Hoc Networks	6
2013	An innovative methodology for detection and quantification of cracks through incorporation of depth perception Mohammad R. Jahanshahi 0001, Sami F. Masri, Curtis Padgett, Gaurav S. Sukhatme	Mach. Vis. Appl.	4
2013	Constrained Interaction and Coordination in Proximity-Limited Multiagent Systems abstract In this paper, we consider the problem of controlling the interactions of a group of mobile agents, subject to a set of topological constraints. Assuming proximity-limited interagent communication, we leverage mobility, unlike prior work, to enable adjacent agents to interact discriminatively, i.e., to actively retain or reject communication links on the basis of constraint satisfaction. Specifically, we propose a distributed scheme that consists of hybrid controllers with discrete switching for link discrimination, coupled with attractive and repulsive potentials fields for mobility control, where constraint violation predicates form the basis for discernment. We analyze the application of constrained interaction to two canonical coordination objectives, i.e., aggregation and dispersion, with maximum and minimum node degree constraints, respectively. For each task, we propose predicates and control potentials, and examine the dynamical properties of the resulting hybrid systems. Simulation results demonstrate the correctness of our proposed methods and the ability of our framework to generate topology-aware coordinated behavior. Ryan K. Williams, Gaurav S. Sukhatme	IEEE Trans. Robotics	2
2012	Opportunistic localization of underwater robots using drifters and boats abstract The paper characterizes the localization performance of an Autonomous Underwater Vehicle (AUV) when it moves in environments where floating drifters or surface vessels are present and can be used for relative localization. In particular, we study how localization performance is affected by parameters e.g. AUV mobility, surface objects density, the available measurements (ranging and/or bearing) and their visibility range. We refer to known techniques for estimation performance evaluation and probabilistic mobility models, and we bring them together to provide a solid numerical analysis for the considered problem. We perform an extensive simulations in different scenarios, and, as a proof of concept, we show how an AUV, equipped with an upward looking sonar, can improve its localization estimate by detecting a surface vessel. Filippo Arrichiello, Hordur Kristinn Heidarsson, Gaurav S. Sukhatme	ICRA	3
2012	Branch and bound for informative path planning abstract We present an optimal algorithm for informative path planning (IPP), using a branch and bound method inspired by feature selection algorithms. The algorithm uses the monotonicity of the objective function to give an objective function-dependent speedup versus brute force search. We present results which suggest that when maximizing variance reduction in a Gaussian process model, the speedup is significant. Jonathan Binney, Gaurav S. Sukhatme	ICRA	2
2012	Using manipulation primitives for brick sorting in clutter abstract This paper explores the idea of manipulation-aided perception and grasping in the context of sorting small objects on a tabletop. We present a robust pipeline that combines perception and manipulation to accurately sort Duplo bricks by color and size. The pipeline uses two simple motion primitives to manipulate the scene in ways that help the robot to improve its perception. This results in the ability to sort cluttered piles of Duplo bricks accurately. We present experimental results on the PR2 robot comparing brick sorting without the aid of manipulation to sorting with manipulation primitives that show the benefits of the latter, particularly as the degree of clutter in the environment increases. Gaurav S. Sukhatme	ICRA	2
2012	Uncertainty-driven view planning for underwater inspection abstract We discuss the problem of inspecting an underwater structure, such as a submerged ship hull, with an autonomous underwater vehicle (AUV). In such scenarios, the goal is to construct an accurate 3D model of the structure and to detect any anomalies (e.g., foreign objects or deformations). We propose a method for constructing 3D meshes from sonar-derived point clouds that provides watertight surfaces, and we introduce uncertainty modeling through non-parametric Bayesian regression. Uncertainty modeling provides novel cost functions for planning the path of the AUV to minimize a metric of inspection performance. We draw connections between the resulting cost functions and submodular optimization, which provides insight into the formal properties of active perception problems. In addition, we present experimental trials that utilize profiling sonar data from ship hull inspection. Geoffrey A. Hollinger, Brendan J. Englot, Franz S. Hover, Urbashi Mitra, Gaurav S. Sukhatme	ICRA	5
2012	Towards improving mission execution for autonomous gliders with an ocean model and kalman filter abstract Effective execution of a planned path by an underwater vehicle is important for proper analysis of the gathered science data, as well as to ensure the safety of the vehicle during the mission. Here, we propose the use of an unscented Kalman filter to aid in determining how the planned mission is executed. Given a set of waypoints that define a planned path and a dicretization of the ocean currents from a regional ocean model, we present an approach to determine the time interval at which the glider should surface to maintain a prescribed tracking error, while also limiting its time on the ocean surface. We assume practical mission parameters provided from previous field trials for the problem set up, and provide the simulated results of the Kalman filter mission planning approach. The results are initially compared to data from prior field experiments in which an autonomous glider executed the same path without pre-planning. Then, the results are validated through field trials with multiple autonomous gliders implementing different surfacing intervals simultaneously while following the same path. Ryan N. Smith, Jonathan Kelly, Gaurav S. Sukhatme	ICRA	3
2012	Probabilistic spatial mapping and curve tracking in distributed multi-agent systems abstract In this paper we consider a probabilistic method for mapping a spatial process over a distributed multi-agent system and a coordinated level curve tracking algorithm for adaptive sampling. As opposed to assuming the independence of spatial features (e.g. an occupancy grid model), we adopt a novel model of spatial dependence based on the grid-structured Markov random field that exploits spatial structure to enhance mapping. The multi-agent Markov random field framework is utilized to distribute the model over the system and to decompose the problem of global inference into local belief propagation problems coupled with neighbor-wise inter-agent message passing. A Lyapunov stable control law for tracking level curves in the plane is derived and a method of gradient and Hessian estimation is presented for applying the control in a probabilistic map of the process. Simulation results over a real-world dataset with the goal of mapping a plume-like oceanographic process demonstrate the efficacy of the proposed algorithms. Scalability and complexity results suggest the feasibility of the approach in realistic multi-agent deployments. Ryan K. Williams, Gaurav S. Sukhatme	ICRA	2
2012	Decentralized Data Fusion and Active Sensing with Mobile Sensors for Modeling and Predicting Spatiotemporal Traffic Phenomena Jie Chen 0027, Kian Hsiang Low, Colin Keng-Yan Tan, Ali Oran, Patrick Jaillet, John M. Dolan, Gaurav S. Sukhatme	UAI	7
2012	Underwater Data Collection Using Robotic Sensor Networks abstract We examine the problem of utilizing an autonomous underwater vehicle (AUV) to collect data from an underwater sensor network. The sensors in the network are equipped with acoustic modems that provide noisy, range-limited communication. The AUV must plan a path that maximizes the information collected while minimizing travel time or fuel expenditure. We propose AUV path planning methods that extend algorithms for variants of the Traveling Salesperson Problem (TSP). While executing a path, the AUV can improve performance by communicating with multiple nodes in the network at once. Such multi-node communication requires a scheduling protocol that is robust to channel variations and interference. To this end, we examine two multiple access protocols for the underwater data collection scenario, one based on deterministic access and another based on random access. We compare the proposed algorithms to baseline strategies through simulated experiments that utilize models derived from experimental test data. Our results demonstrate that properly designed communication models and scheduling protocols are essential for choosing the appropriate path planning algorithms for data collection. Geoffrey A. Hollinger, Sunav Choudhary, Parastoo Qarabaqi, Chris Murphy, Urbashi Mitra, Gaurav S. Sukhatme, Milica Stojanovic, Hanumant Singh, Franz S. Hover	IEEE J. Sel. Areas Commun.	6
2011	A Data-Driven Movement Model for Single Cellphone-Based Indoor Positioning abstract Indoor localization is a promising area with applications in in-home monitoring and tracking. Fingerprinting and propagation model-based WiFi localization techniques have limited spatial resolution because of grid or graph-based representations. An alternative is to incorporate dynamics models based on real-time sensing of human movement and fuse these with WiFi measurements. We present a data-driven dynamic model that tracks the inherent periodicity in walking and converts this representation into velocity. This model however is prone to drift. We correct this drift with WiFi measurements to obtain a combined position estimate. Our approach records and fuses human body movement with WiFi positioning using a single mobile phone. We characterize the movement model to obtain an estimate of error predictions. The movement model showed a best case average RMS prediction error of .25 m/s. We also present a preliminary study characterizing combined system performance across straight line and L-shaped trajectories. The framework showed lower errors across the L-shaped trajectories (mean error = 4.8 m using movement sensing versus a mean error = 6 m without movement sensing) because of the ability assess the validity of a WiFi measurement. Higher errors were observed across the straight line trajectory due to imprecise trajectories. Harshvardhan Vathsangam, Anupam Tulsyan, Gaurav S. Sukhatme	BSN	3
2011	Cooperative control of autonomous surface vehicles for oil skimming and cleanup abstract Oil skimmers towed by two vehicles have been widely used for skimming of oil on the water surface. In this paper, we address the cooperative control of two autonomous surface vehicles for oil skimming and cleanings. We model the skimmer as a flexible, floating rope of constant length as well as discrete segmented model. We derive the equations governing the rope dynamics from first principles and demonstrate their application through simulations. We have performed field experiments with two autonomous surface vehicles that substantiate the proposed model and provides estimates of constants underlying the model. We propose a method for controlling the shape of the rope, and derive the conditions that maximize skimming efficiency. Subhrajit Bhattacharya, Hordur Kristinn Heidarsson, Gaurav S. Sukhatme, Vijay Kumar 0001	ICRA	3
2011	Obstacle detection and avoidance for an Autonomous Surface Vehicle using a profiling sonar abstract We present an experimental study of a mechanically scanned profiling sonar for Autonomous Surface Vehicle (ASV) obstacle detection and avoidance. We extract potential obstacles from echo returns and suggest a scanning strategy for sonar in this application. We demonstrate with simulations (driven by data collected in the field) the potential for an ASV to rely solely on sonar data to navigate and avoid obstacles in a lake and harbor environment. Hordur Kristinn Heidarsson, Gaurav S. Sukhatme	ICRA	2
2011	Distributed coordination and data fusion for underwater search abstract This paper presents coordination and data fusion methods for teams of vehicles performing target search tasks without guaranteed communication. A fully distributed team planning algorithm is proposed that utilizes limited shared information as it becomes available, and data fusion techniques are introduced for merging estimates of the target's position from vehicles that regain contact after long periods of time. The proposed data fusion techniques are shown to avoid overcounting information, which ensures that combining data from different vehicles will not decrease the performance of the search. Motivated by the underwater search domain, a realistic underwater acoustic communication channel is used to determine the probability of successful data transfer between two locations. The channel model is integrated into a simulation of multiple autonomous vehicles in both open ocean and harbor search scenarios. The simulated experiments demonstrate that distributed coordination with limited communication significantly improves team performance versus prior techniques that continually maintain connectivity. Geoffrey A. Hollinger, Srinivas Yerramalli, Sanjiv Singh, Urbashi Mitra, Gaurav S. Sukhatme	ICRA	5
2011	Mission design for compressive sensing with mobile robots abstract This paper considers mission design strategies for mobile robots whose task is to perform spatial sampling of a static environmental field, in the framework of compressive sensing. According to this theory, we can reconstruct compressible fields using O(log n) nonadaptive measurements (where n is the number of sites of the spatial domain), in a basis that is "in coherent" to the representation basis [1]; random uncorrelated measurements satisfy this incoherence requirement. Because an autonomous vehicle is kinematically constrained and has finite energy and communication resources, it is an open question how to best design missions for CS reconstruction. We compare a two-dimensional random walk, a TSP approximation to pass through random points, and a randomized boustrophedon (lawnmower) strategy. Not unexpectedly, all three approaches can yield comparable reconstruction performance if the planning horizons are long enough; if planning occurs only over short time scales, the random walk will have an advantage. Robert Hummel, Sameera Poduri, Franz S. Hover, Urbashi Mitra, Gaurav S. Sukhatme	ICRA	5
2011	Simultaneous mapping and stereo extrinsic parameter calibration using GPS measurements abstract Stereo vision is useful for a variety of robotics tasks, such as navigation and obstacle avoidance. However, recovery of valid range data from stereo depends on accurate calibration of the extrinsic parameters of the stereo rig, i.e., the 6-DOF transform between the left and right cameras. Stereo self calibration is possible, but, without additional information, the absolute scale of the stereo baseline cannot be determined. In this paper, we formulate stereo extrinsic parameter calibration as a batch maximum likelihood estimation problem, and use GPS measurements to establish the scale of both the scene and the stereo baseline. Our approach is similar to photogrammetric bundle adjustment, and closely related to many structure from motion algorithms. We present results from simulation experiments using a range of GPS accuracy levels; these accuracies are achievable by varying grades of commercially-available receivers. We then validate the algorithm using stereo and GPS data acquired from a moving vehicle. Our results indicate that the approach is promising. Jonathan Kelly, Larry H. Matthies, Gaurav S. Sukhatme	ICRA	3
2011	Persistent ocean monitoring with underwater gliders: Towards accurate reconstruction of dynamic ocean processes abstract This paper proposes a path planning algorithm and a velocity control algorithm for underwater gliders to persistently monitor a patch of ocean. The algorithms address a pressing need among ocean scientists to collect high-value data for studying ocean events of scientific and environmental interest, such as the occurrence of harmful algal blooms. The path planner optimizes a cost function that blends two competing factors: it maximizes the information value of the path, while minimizing the deviation from the path due to ocean currents. The speed control algorithm then optimizes the speed along the planned path so that higher resolution samples are collected in areas of higher information value. The resulting paths are closed circuits that can be repeatedly traversed to collect long term ocean data in dynamic environments. The algorithms were tested during sea trials on an underwater glider operating off the coast of southern California over the course of several weeks. The results show significant improvements in data resolution and path reliability compared to a sampling path that is typically used in the region. Ryan N. Smith, Mac Schwager, Stephen L. Smith 0001, Daniela Rus, Gaurav S. Sukhatme	ICRA	5
2011	Towards autonomous wireless backbone deployment in highly-obstructed environments abstract In a setting that lacks infrastructure e.g., urban search and rescue, a team of networked mobile robots can provide a communication substrate by acting as routers in a wireless mesh network. We study the problem of determining the minimum number of robots, and how to position them, so that all clients using the resulting robotic network are connected and all network links satisfy minimum rate requirements. The key challenge we address is that in an environment with obstacles the strength of a wireless link is a non-monotonic function of the distance between the link end-points. Our approach to the problem is based on virtual potential fields. Clients and environmental obstacles are modeled as virtual charged particles exerting virtual forces on the robots. We validate our algorithm with physical robots in an indoor environment and demonstrate that we are able to get feasible solutions. Marcos A. M. Vieira, Ramesh Govindan, Gaurav S. Sukhatme	ICRA	3
2011	Towards mixed-initiative, multi-robot field experiments: Design, deployment, and lessons learned abstract With the advent of Autonomous Underwater Vehicles (AUVs) and other mobile platforms, marine robotics have had substantial impact on the oceanographic sciences. These systems have allowed scientists to collect data over temporal and spatial scales that would be logistically impossible or prohibitively expensive using traditional ship-based measurement techniques. Increased dependence of scientists on such robots has permeated scientific data gathering with future field campaigns involving these platforms as well as on entire infrastructure of people, processes and software, on shore and at sea. Recent field experiments carried out with a number of surface and underwater platforms give clues to how these technologies are coalescing and need to work together. We highlight one such confluence and describe a future trajectory of needs and desires for field experiments with autonomous marine robotic platforms. Our 2010 inter-disciplinary experiment in the Monterey Bay involved multiple platforms and collaborators with diverse science goals. One important goal was to enable situational awareness, planning and collaboration before, during and after this large-scale collaborative exercise. We present the overall view of the experiment and describe an important shore-side component, the Oceanographic Decision Support System (ODSS), its impact and future directions leveraging such technologies for field experiments. Jnaneshwar Das, Thom Maughan, Mike McCann, Mike Godin, Tom O'Reilly, Monique Messie, Fred Bahr, Kevin Gomes, Frédéric Py, James G. Bellingham, Gaurav S. Sukhatme, Kanna Rajan	IROS	11
2011	Obstacle detection from overhead imagery using self-supervised learning for Autonomous Surface Vehicles abstract We describe a technique for an Autonomous Surface Vehicle (ASV) to learn an obstacle map by classifying overhead imagery. Classification labels are supplied by a front-facing sonar, mounted under the water line on the ASV. We use aerial imagery from two online sources for each of two water bodies (a small lake and a harbor) and train classifiers using features generated from each image source separately, followed by combining their output. Data collected using a sonar mounted on the ASV were used to generate the labels in the experimental study. The results show that we are able to generate accurate obstacle maps well-suited for ASV navigation. Hordur Kristinn Heidarsson, Gaurav S. Sukhatme	IROS	2
2011	Autonomous data collection from underwater sensor networks using acoustic communication abstract We examine the problem of planning paths for an autonomous underwater vehicle (AUV) to collect data from an underwater sensor network. The sensors in the network are equipped with acoustic modems that provide noisy, range-limited communication. The AUV must plan a path that maximizes the information collected while minimizing travel time or fuel expenditure. This problem is closely related to the classical Traveling Salesperson Problem (TSP), but differs in that data from a particular sensor has a probability of being collected depending on the quality of communication. We propose methods for solving this problem by extending approximation algorithms for variants of TSP, and we compare our proposed algorithms to baseline strategies through simulated experiments with varying levels of communication quality. Our simulations utilize a realistic model of acoustic communication to determine the probability of acquiring data from each sensor. The results demonstrate that planning the tour for the entire network while exploiting the communication model during planning improves performance versus myopic methods. Geoffrey A. Hollinger, Urbashi Mitra, Gaurav S. Sukhatme	IROS	3
2011	Toward risk aware mission planning for Autonomous Underwater Vehicles abstract Long range and high endurance Autonomous Underwater Vehicles such as gliders enable sustained oceanographic sampling at larger time-scales and much lower operational costs compared to traditional ship-based sampling methods. While most path-planning methods for AUVs optimize paths with respect to efficiency, obstacle avoidance, and control they do not explicitly address the issue of finding the safest possible path when considering risks such as shipping traffic and bathymetry. In coastal regions with high shipping traffic, reducing collision risk at the path planning stage, at the expense of efficiency, is a worthwhile trade-off. We propose a method of building risk maps using historical data from the Automated Information System. These are used to plan minimum risk paths between a specified start and goal location, while avoiding obstacles, using an algorithm based on A* search. Our planner incorporates the uncertainty in dead-reckoning without explicitly considering the effect of ocean currents. We compare the relative risk of paths produced by our method when compared to a shortest-path planner which does not take risk into account, and show that our methods performs significantly better, while producing competitive paths lengths. Arvind Pereira, Jonathan Binney, Burton H. Jones, Matthew Ragan, Gaurav S. Sukhatme	IROS	5
2011	Cooperative multi-agent inference over grid structured Markov random fields abstract In this work we investigate cooperative inference in multi-agent systems where uncertainty is modeled by the grid structured pairwise Markov random field. A framework is proposed, which we term the multi-agent Markov random field, that decomposes the global inference problem into inter-agent belief exchanges over a hypertree topology and local intra-agent inference problems. Due to the exponential complexity of exact inference, we propose a loopy belief propagation algorithm for approximate inference over appropriately formed local generalized cluster graphs. Both synchronous and intelligent message passing are considered and a grid scale-invariant scheme based on the notion of regions of influence in a cluster graph is presented. The algorithms are simulated over a grid workspace with a team of virtual Autonomous Surface Vehicles (ASVs), with the goal of spatial plume detection in oceanographic data captured from the Moderate Resolution Imaging Spectroradiometer (MODIS) instrument. We show that while the exact method produces predictably accurate and smooth grid maps, the approximate method competes well in terms of plume detection rate with the region of influence message passing scheme excelling over large tasks due to a lack of dependence on grid size. Ryan K. Williams, Gaurav S. Sukhatme	IROS	2
2011	Active Classification: Theory and Application to Underwater Inspection Geoffrey A. Hollinger, Urbashi Mitra, Gaurav S. Sukhatme	ISRR	3
2010	Observability analysis of relative localization for AUVs based on ranging and depth measurements abstract The paper studies the observability properties of the relative localization of two Autonomous Underwater Vehicles (AUVs) equipped with depth sensors, linear/angular velocity sensors, and communication devices with range measurement. The conditions that ensure observability of the linearized model and locally weak observability of the nonlinear system are derived. An Extended Kalman Filter is then designed aimed at estimating the relative position between two AUVs. Simulations in 3D and reconstruction from experimental data in 2D provide a numerical validation of the analysis. Gianluca Antonelli, Filippo Arrichiello, Stefano Chiaverini, Gaurav S. Sukhatme	ICRA	4
2010	Cooperative caging using autonomous aquatic surface vehicles abstract We present a study on the use of cooperative robots to execute a caging mission on the water's surface. In particular, we consider the problem of using two robotic boats (under-actuated autonomous surface vessels) connected with a floating rope, to `capture' a floating object from a known location on the water's surface and 'shepherd' it to a designated position. This paper focuses on the cooperative control strategy of the two vessels. Each vessel's behavior is governed by a supervisor software module that handles the communication with the other vessel and controls all elementary tasks that compose the overall mission. The elementary tasks, specifically developed for under-actuated vessels, are arranged by priority, and merged using a behavior-based approach, namely the Null-Space based Behavioral control. The proposed technique is validated by field experiments with two autonomous robotic boats on the surface of a lake. Filippo Arrichiello, Hordur Kristinn Heidarsson, Stefano Chiaverini, Gaurav S. Sukhatme	ICRA	4
2010	Informative path planning for an autonomous underwater vehicle abstract We present a path planning method for autonomous underwater vehicles in order to maximize mutual information. We adapt a method previously used for surface vehicles, and extend it to deal with the unique characteristics of underwater vehicles. We show how to generate near-optimal paths while ensuring that the vehicle stays out of high-traffic areas during predesignated time intervals. In our objective function we explicitly account for the fact that underwater vehicles typically take measurements while moving, and that they do not have the ability to communicate until they resurface. We present field results from ocean trials on planning paths for a specific AUV, an underwater glider. Jonathan Binney, Andreas Krause 0001, Gaurav S. Sukhatme	ICRA	3
2010	Weighted barrier functions for computation of force distributions with friction cone constraints abstract We present a novel Weighted Barrier Function (WBF) method of efficiently computing optimal grasping force distributions for multifingered hands. Second-order conic friction constraints are not linearized, as in many previous works. The force distributions are smooth and rapidly computable, and they enable flexibility in selecting between firm, stable grasps or looser, more efficient grasps. Furthermore, fingers can be disengaged and re-engaged in a smooth manner, which is a critical capability for a large number of manipulation tasks. We present efficient solution methods that do not incur the increased computational complexity associated with solving the Semi-Definite Programming formulations presented in previous works. We present results from static and dynamic simulations which demonstrate the flexibility and computational efficiency associated with WBF force distributions. Per Henrik Borgstrom, Maxim A. Batalin, Gaurav S. Sukhatme, William J. Kaiser	ICRA	3
2010	Towards marine bloom trajectory prediction for AUV mission planning abstract This paper presents an oceanographic toolchain that can be used to generate multi-vehicle robotic surveys for large-scale dynamic features in the coastal ocean. Our science application targets Harmful Algal Blooms (HABs) which have significant societal impact to coastal communities yet are poorly understood ecologically. Bloom patches can be large spatially (in kms) and unpredictable in their extent. To understand their ecology, we need to be able to bring back water samples from the `right' places and times for lab analysis. In doing so, we target hotspots representative of intense biogeochemical activity for such sampling. Our approach uses remote sensing data to detect such hotspots using ocean color as a proxy, and advectively projects these patches spatio-temporally using surface current data from HF Radar stations. Experiments with satellite and Radar data sets are promising for large, coherent blooms. We show how these predictions can be used to select an appropriate sampling trajectory for an AUV. Jnaneshwar Das, Kanna Rajan, Sergey Frolov, Frédéric Py, John P. Ryan 0001, David A. Caron, Gaurav S. Sukhatme	ICRA	7
2010	Autonomous Underwater Vehicle trajectory design coupled with predictive ocean models: A case study abstract Data collection using Autonomous Underwater Vehicles (AUVs) is increasing in importance within the oceanographic research community. Contrary to traditional moored or static platforms, mobile sensors require intelligent planning strategies to maneuver through the ocean. However, the ability to navigate to high-value locations and collect data with specific scientific merit is worth the planning efforts. In this study, we examine the use of ocean model predictions to determine the locations to be visited by an AUV, and aid in planning the trajectory that the vehicle executes during the sampling mission. The objectives are: a) to provide near-real time, in situ measurements to a large-scale ocean model to increase the skill of future predictions, and b) to utilize ocean model predictions as a component in an end-to-end autonomous prediction and tasking system for aquatic, mobile sensor networks. We present an algorithm designed to generate paths for AUVs to track a dynamically evolving ocean feature utilizing ocean model predictions. This builds on previous work in this area by incorporating the predicted current velocities into the path planning to assist in solving the 3-D motion planning problem of steering an AUV between two selected locations. We present simulation results for tracking a fresh water plume by use of our algorithm. Additionally, we present experimental results from field trials that test the skill of the model used as well as the incorporation of the model predictions into an AUV trajectory planner. These results indicate a modest, but measurable, improvement in surfacing error when the model predictions are incorporated into the planner. Ryan N. Smith, Arvind Pereira, Yi Chao, Peggy Li, David A. Caron, Burton H. Jones, Gaurav S. Sukhatme	ICRA	7
2010	An architecture-driven software mobility framework Sam Malek, George Edwards, Yuriy Brun, Hossein Tajalli, Joshua Garcia, Ivo Krka, Nenad Medvidovic, Marija Mikic-Rakic, Gaurav S. Sukhatme	J. Syst. Softw.	9
2009	3D tree reconstruction from laser range data abstract We present a method for reconstructing 3D models of tree branch structure from laser range data. Our approach is probabilistic, and uses general knowledge of tree structure to guide an iterative reconstruction process. Our goal is to recover parameters such as branch locations, angles, radii, and lengths, as well as connectivity information between branches. These parameters can then be fed into functional-structural plant models to study the relationships between the structure of a plant, its environment, and its internal biology. In this paper we present an algorithm for finding these parameters, and results on both simulated and real datasets. Jonathan Binney, Gaurav S. Sukhatme	ICRA	2
2009	Relative bearing estimation from commodity radios abstract Relative bearing between robots is important in applications like pursuit-evasion [11] and SLAM [7]. This is also true in sensor networks, where the bearing of one sensor node relative to another has been used for localization [5], [18], [20] and topology control [14], [21], [6]. Most systems use dedicated sensors like an IR array or a camera to obtain relative bearing. We study the use of radio signal strength (RSS) in commodity radios for obtaining relative bearing. We show that by using the robot's mobility, commodity radios can be used to obtain coarse relative bearing. This measurement can be used for a suite of applications that do not require very precise bearing measurement. We analyze signal strength variations in simulation and experiment and also show an algorithm that uses this coarse bearing computation in a practical setting. Karthik Dantu, Prakhar Goyal, Gaurav S. Sukhatme	ICRA	3
2009	A robotic sentinel for benthic sampling along a transect abstract This paper presents the design of a novel robotic system capable of long-term benthic sampling along a transect. The robot is built to traverse back and forth along a mechanical guide-rail at the bottom of a water body. We present results from localization tests with the robot in a laboratory tank and a shallow swimming pool. A pilot deployment was made at a marina to accurately observe the rate and direction of water flow across a section of the marina inlet. Results from the experiment demonstrate the potential of this platform for monitoring various aquatic phenomena of interest. Jnaneshwar Das, Gaurav S. Sukhatme	ICRA	2
2009	Distributed coverage control for mobile sensors with location-dependent sensing models abstract This paper addresses the problem of coverage control of a network of mobile sensors. In the current literature, this is commonly formulated as a locational optimization problem under the assumption that sensing performance is independent of the locations of sensors. We extend this work to a more general framework where the sensor model is location-dependent. We propose a distributed control law and coordination algorithm. If the global sensing performance function is known a priori, we prove that the algorithm is guaranteed to converge. To validate this algorithm, we conduct experiments with indoor and outdoor deployments of Cyclops cameras and model its sensing performance. This model is used to simulate deployments on 1D pathways and study the coverage obtained. We also examine the coverage in the case when the global sensing function is not known and is estimated in an online fashion. Ajay Deshpande, Sameera Poduri, Daniela Rus, Gaurav S. Sukhatme	ICRA	4
2009	Collective transport of robots: Coherent, minimalist multi-robot leader-following abstract We study the collective transport of robots (CTR) problem. A large number of commodity mobile robots are to be moved from one location to another by a single operator. Joysticking each one or carrying them physically is impractical. None of the robots are particularly sophisticated in their ability to plan or reason. Prior work on flocking and formation control has addressed the transport of a robot group that maintains its integrity by explicitly controlling coherence. We show how flocking emerges as a consequence of each robot contending for space near the human operator. A coherent flock can be made to follow a leader in this manner thereby solving the CTR problem. We also present the design of a hand-worn IMU-based gesture interface which allows the human operator to issue simple commands to the group. A preliminary experimental evaluation of the system shows robust CTR with different leader behaviors. Jnaneshwar Das, Marcos A. M. Vieira, Hordur Kristinn Heidarsson, Harshvardhan Vathsangam, Gaurav S. Sukhatme	IROS	6
2009	Using Local Geometry for Tunable Topology Control in Sensor Networks abstract Neighbor-Every-Theta (NET) graphs are such that each node has at least one neighbor in every theta angle sector of its communication range. We show that for thetas < pi, NET graphs are guaranteed to have an edge-connectivity of at least floor (2pi)/thetas, even with an irregular communication range. Our main contribution is to show how this family of graphs can achieve tunable topology control based on a single parameter thetas. Since the required condition is purely local and geometric, it allows for distributed topology control. For a static network scenario, a power control algorithm based on the NET condition is developed for obtaining k-connected topologies and shown to be significantly efficient compared to existing schemes. In controlled deployment of a mobile network, control over positions of nodes can be leveraged for constructing NET graphs with desired levels of network connectivity and sensing coverage. To establish this, we develop a potential fields based distributed controller and present simulation results for a large network of robots. Lastly, we extend NET graphs to 3D and provide an efficient algorithm to check for the NET condition at each node. This algorithm can be used for implementing generic topology control algorithms in 3D. Sameera Poduri, Sundeep Pattem, Bhaskar Krishnamachari, Gaurav S. Sukhatme	IEEE Trans. Mob. Comput.	4
2009	Design and Implementation of NIMS3D, a 3-D Cabled Robot for Actuated Sensing Applications abstract We present NIMS3D, a novel 3-D cabled robot for actuated sensing applications. We provide a brief overview of the main hardware components. Next, we describe installation procedures, including novel calibration methods, that enable rapid in-field deployability for nonexpert end users, and provide simulations and experimental results to highlight their effectiveness. Kinematic and dynamic analysis of the system are provided, followed by a description of control methods. We provide experimental results that illustrate tracking of linear and nonlinear paths by NIMS3D. Thereafter, we briefly present an example of an actuated sensing task performed by the system. Finally, we describe methods of improving energy efficiency by leveraging nonlinear trajectories and energy-optimal tension distributions. Experimental and simulated results show that energy efficiency can be improved significantly by using optimized parabolic trajectories. Furthermore, we provide simulation results that demonstrate improved efficiency enabled by optimal, least norm tension distributions. Per Henrik Borgstrom, Nils Peter Borgstrom, Michael J. Stealey, Brett L. Jordan, Gaurav S. Sukhatme, Maxim A. Batalin, William J. Kaiser	IEEE Trans. Robotics	5
2009	NIMS-PL: A Cable-Driven Robot With Self-Calibration Capabilities abstract We present the Networked InfoMechanical System for Planar Translation, which is a novel two-degree-of-freedom (2-DOF) cable-driven robot with self-calibration and online drift-correction capabilities. This system is intended for actuated sensing applications in aquatic environments. The actuation redundancy resulting from in-plane translation driven by four cables results in an infinite set of tension distributions, thus requiring real-time computation of optimal tension distributions. To this end, we have implemented a highly efficient, iterative linear programming solver, which requires a very small number of iterations to converge to the optimal value. In addition, two novel self-calibration methods have been developed that leverage the robot's actuation redundancy. The first uses an incremental displacement, or jitter method, whereas the second uses variations in cable tensions to determine end-effector location. We also propose a novel least-squares drift-detection algorithm, which enables the robot to detect long-term drift. Combined with self-calibration capabilities, this drift-monitoring algorithm enables long-term autonomous operation. To verify the performance of our algorithms, we have performed extensive experiments in simulation and on a real system. Per Henrik Borgstrom, Brett L. Jordan, Bengt J. Borgstrom, Michael J. Stealey, Gaurav S. Sukhatme, Maxim A. Batalin, William J. Kaiser	IEEE Trans. Robotics	5
2009	Rapid Computation of Optimally Safe Tension Distributions for Parallel Cable-Driven Robots abstract In this paper, we present a novel linear-program formulation that yields "optimally safe" (OS) tension distributions in parallel cable-driven robots by the introduction of a slack variable. The slack variable also enables explicit computation of a near-optimal, feasible starting point. This, in turn, enables rapid computation of the OS tension distributions. The formulation also contains a parameter that can be used to steer cable tensions toward desired regions of operation. We present static results from two simulated robotic systems that demonstrate the ability of our formulation to avoid tension limits. Simulated execution of highly dynamic trajectories on both systems demonstrates rapid-computation abilities. Furthermore, we present experimental results from a real robotic system that further validate the importance of safe tension distributions. Per Henrik Borgstrom, Brett L. Jordan, Gaurav S. Sukhatme, Maxim A. Batalin, William J. Kaiser	IEEE Trans. Robotics	3
2008	Generation of energy efficient trajectories for NIMS3D, a three-dimensional cabled robot abstract In this paper we describe an algorithm to generate energy efficient trajectories for NIMS3D, a three-dimensional cabled robotic platform. Optimized parabolic paths are used to exploit the relatively low I2R loss associated with operation in lower regions of the workspace. Trajectory optimization is sufficiently fast to enable real time operation. Experimental results on a physical system for a three cable deployment show substantial reductions in energy consumption as compared to linear trajectories. Per Henrik Borgstrom, Nils Peter Borgstrom, Michael J. Stealey, Brett L. Jordan, Gaurav S. Sukhatme, Maxim A. Batalin, William J. Kaiser	ICRA	5
2008	Towards spatial and semantic mapping in aquatic environments abstract High fidelity data acquisition of dynamic spatiotemporal phenomena for aquatic environmental research suggests the use of actuated sensors. Furthermore, characterization of the floor in aquatic environments is beneficial for environmental science, as well as can be applied to robot localization. The NIMS AQ cable robot platform is designed to meet these requirements and satisfy the constraints of large scale, in-field deployments. In addition to a set of water quality sensors it also carries an ultra-miniature side-scan sonar. In this paper we show the development of methods for autonomous range detection, spatial and semantic mapping in underwater environments. These methods are demonstrated to be important for future developments including localization, navigation, and path planning, particularly for 3D mobility. Experiments have been performed in both controlled environments and a lake environment and results are discussed. Victor Chen 0001, Maxim A. Batalin, William J. Kaiser, Gaurav S. Sukhatme	ICRA	4
2008	Energy based path planning for a novel cabled robotic system abstract Cabled robotic systems have been used for a diverse set of applications such as environmental sensing, search and rescue, sports and entertainment and air vehicle simulators. In this paper, we introduce a new cabled robot- Networked Info Mechanical System for Planar actuation (NIMS-PL), with energy profiling capabilities. Accurate energy measurements supported by NIMS-PL enable path planning that optimizes the robotpsilas path subject to an upper bound on energy consumption. We performed extensive empirical validation of the optimized path planning approach in simulation using an environmental sensing application as an example. We also validated the simulation results using NIMS-PL, demonstrating significant improvements in the sensing task when accounting with accurate energy measurements as opposed to Euclidean distance, which is typically used for modeling energy spent in path traversal. Per Henrik Borgstrom, Amarjeet Singh 0001, Brett L. Jordan, Gaurav S. Sukhatme, Maxim A. Batalin, William J. Kaiser	IROS	4
2008	An experimental study of station keeping on an underactuated ASV abstract Dynamic positioning is an important application for marine vehicles that do not have the luxury of anchoring or mooring themselves. Such vehicles are usually large and have arrays of thrusters that allow for controllability in the sway as well as the surge and yaw axes. Most smaller boats however, are underactuated and do not possess control in the sway direction. This makes the control problem significantly more challenging. We address the station keeping problem for a small autonomous surface vehicle (ASV) with significant windage. The vehicle is required to hold station at a given position. We describe the design of a weighted controller that uses wind feed-forward to complement a line-of-sight guidance controller to achieve satisfactory performance under slow-varying moderate wind conditions. We test the control system in simulation and in field trials with a twin-propeller ASV. Experiments show that the controller works very well in moderate wind conditions allowing the ASV to keep station with a position error of approximately one vehicle length. Arvind Pereira, Jnaneshwar Das, Gaurav S. Sukhatme	IROS	3
2008	Semantic Mapping Using Mobile Robots abstract Robotic mapping is the process of automatically constructing an environment representation using mobile robots. We address the problem of semantic mapping, which consists of using mobile robots to create maps that represent not only metric occupancy but also other properties of the environment. Specifically, we develop techniques to build maps that represent activity and navigability of the environment. Our approach to semantic mapping is to combine machine learning techniques with standard mapping algorithms. Supervised learning methods are used to automatically associate properties of space to the desired classification patterns. We present two methods, the first based on hidden Markov models and the second on support vector machines. Both approaches have been tested and experimentally validated in two problem domains: terrain mapping and activity-based mapping. Denis F. Wolf, Gaurav S. Sukhatme	IEEE Trans. Robotics	2
2007	Detecting and Tracking Level Sets of Scalar Fields using a Robotic Sensor Network abstract We introduce an algorithm which detects and traces a specified level set of a scalar field (a contour) on a plane. A network of static sensor nodes with limited communication and processing are deployed in a planar environment along with a mobile node which can both sense and move. As the mobile node moves through the environment, it computes the local spatial gradient of the field by communicating with its immediate neighbors in the static sensor network. The algorithm causes the mobile node to perform gradient descent on the scalar field till it arrives at a location on the desired contour. From this point onwards, the algorithm drives the mobile node to trace the desired contour without departing from it. Experiments in simulation indicate that the required contour is found with reasonable accuracy (between 80-90%) for networks with node degree of greater than or equal to six. Our results also indicate that the paths generated by our algorithm are near-optimal in terms of the distance traversed by the mobile node. Our preliminary experimental results with a physical robot show that our algorithm is feasible. Karthik Dantu, Gaurav S. Sukhatme	ICRA	2
2007	Optimal Control Using Nonholonomic Integrators abstract This paper addresses the optimal control of nonholonomic systems through provably correct discretization of the system dynamics. The essence of the approach lies in the discretization of the Lagrange-d'Alembert principle which results in a set of forced discrete Euler-Lagrange equations and discrete nonholonomic constraints that serve as equality constraints for the optimization of a given cost functional. The method is used to investigate optimal trajectories of wheeled robots. Marin Kobilarov, Gaurav S. Sukhatme	ICRA	2
2007	Latency Analysis of Coalescence for Robot Groups abstract Coalescence is the problem of isolated mobile robots independently searching for peers with the goal of forming a single connected network. This is important because communication is a necessary requirement for several collaborative robot tasks. In this paper, we consider a scenario where the robots do not have any information about the environment or positions of other robots and perform a random walk search. We show through probabilistic analysis that as the number of isolated robots N increases, the expected coalescence time decreases as 1/radicN. Simulations results are presented to validate this analysis. Sameera Poduri, Gaurav S. Sukhatme	ICRA	2
2007	Landing a Helicopter on a Moving Target abstract We present the design of an optimal trajectory controller for landing a helicopter on a moving target. The trajectory planner is based on the variational Hamiltonian and Euler-Lagrange equations. We use a kinematic model of the helicopter to derive an optimal controller that is able to track an arbitrarily moving target and then land on it. Simulations are shown to verify the performance of the optimal trajectory controller. Data from real flight trials is presented to validate the inputs obtained from the trajectory planner to track a desired trajectory. We present initial trials in simulation for landing the helicopter autonomously on a moving target. Srikanth Saripalli, Gaurav S. Sukhatme	ICRA	2
2007	Adaptive Sampling for Estimating a Scalar Field using a Robotic Boat and a Sensor Network abstract This paper introduces an adaptive sampling algorithm for a mobile sensor network to estimate a scalar field. The sensor network consists of static nodes and one mobile robot. The static nodes are able to take sensor readings continuously in place, while the mobile robot is able to move and sample at multiple locations. The measurements from the robot and the static nodes are used to reconstruct an underlying scalar field. The algorithm presented in this paper accepts the measurements made by the static nodes as inputs and computes a path for the mobile robot which minimizes the integrated mean square error of the reconstructed field subject to the constraint that the robot has limited energy. We assume that the field does not change when robot is taking samples. In addition to simulations, we have validated the algorithm on a robotic boat and a system of static buoys operating in a lake over several km of traversed distance while reconstructing the temperature field of the lake surface Bin Zhang 0010, Gaurav S. Sukhatme	ICRA	2
2007	Identifying and Addressing Uncertainty in Architecture-Level Software Reliability Modeling abstract Assessing reliability at early stages of software development, such as at the level of software architecture, is desirable and can provide a cost-effective way of improving a software system's quality. However, predicting a component's reliability at the architectural level is challenging because of uncertainties associated with the system and its individual components due to the lack of information. This paper discusses representative uncertainties which we have identified at the level of a system's components, and illustrates how to represent them in our reliability modeling framework. Our preliminary evaluation indicates promising results in our framework's ability to handle such uncertainties. Leslie Cheung, Leana Golubchik, Nenad Medvidovic, Gaurav S. Sukhatme	IPDPS	4
2007	Discrete trajectory control algorithms for NIMS3D, an autonomous underconstrained three-dimensional cabled robot abstract In this paper we present algorithms that enable precise trajectory control of NIMS3D, an underconstrained, three-dimensional cabled robot intended for use in actuated sensing. We begin by offering a brief system overview and then describe methods to determine the range of operation of the robot. Next, a discrete-time model of the system is presented. Thereafter, we present an online algorithm for modeling motor behavior. The majority of the paper is dedicated to describing three feedback control laws used to enable accurate trajectory tracking for both linear and non-linear motion profiles. We present experimental results that highlight the strengths and weaknesses of these mechanisms and conclude by offering a series of future plans for NIMS3D. Per Henrik Borgstrom, Nils Peter Borgstrom, Michael J. Stealey, Brett L. Jordan, Gaurav S. Sukhatme, Maxim A. Batalin, William J. Kaiser	IROS	5
2007	Experiments in robotic boat localization abstract We are motivated by the prospect of automating microbial observing systems. To this end we have designed and built a robotic boat as part of a sensor network for monitoring aquatic environments. In this paper, we describe a dynamic model of the boat, an algorithm for estimating its location by integrating various sensor inputs, a controller for waypoint following and extensive field experiments (over 10 km aggregate) to validate each of these. We test the localization accuracy in different sensing regimes as a prelude to accommodating sensing failures. Amit Dhariwal, Gaurav S. Sukhatme	IROS	2
2007	Reconfiguration methods for mobile sensor networks abstract Motion may be used in sensor networks to change the network configuration for improving the sensing performance. We consider the problem of controlling motion in a distributed manner for a mobile sensor network for a specific form of motion capability. Mobility itself may have a high resource overhead, hence we exploit motility , a constrained form of mobility, which has very low overheads but provides significant reconfiguration potential. We present an architecture that allows each node in the network to learn the medium and phenomenon characteristics. We describe a quantitative metric for sensing performance that is concretely tied to real sensor and medium characteristics, rather than assuming an abstract range based model. The problem of determining the desirable network configuration is expressed as an optimization of this metric. We present a distributed optimization algorithm which computes a desirable network configuration, and adapts it to environmental changes. The relationship of the proposed algorithm to simulated annealing and incremental subgradient descent based methods is discussed. A key property of our algorithm is that convergence to a desirable configuration can be proved even though no global coordination is involved. A network protocol to implement this algorithm is discussed, followed by simulations and experiments on a laboratory test bed. Aman Kansal, William J. Kaiser, Gregory J. Pottie, Mani Srivastava 0001, Gaurav S. Sukhatme	ACM Trans. Sens. Networks	5
2007	The Design and Analysis of an Efficient Local Algorithm for Coverage and Exploration Based on Sensor Network Deployment abstract We present the design and theoretical analysis of a novel algorithm termed least recently visited (LRV). LRV efficiently and simultaneously solves the problems of coverage, exploration, and sensor network deployment. The basic premise behind the algorithm is that a robot carries network nodes as a payload, and in the process of moving around, emplaces the nodes into the environment based on certain local criteria. In turn, the nodes emit navigation directions for the robot as it goes by. Nodes recommend directions least recently visited by the robot, hence, the name LRV. We formally establish the following two properties: 1) LRV is complete on graphs and 2) LRV is optimal on trees. We present experimental conjectures for LRV on regular square and cube lattice graphs and compare its performance empirically to other graph exploration algorithms. We study the effects of the order of the exploration and show on a square lattice that with an appropriately chosen order, LRV performs optimally. Finally, we discuss the implementation of LRV in simulation and in real hardware. Maxim A. Batalin, Gaurav S. Sukhatme	IEEE Trans. Robotics	2
2006	People Tracking and Following with Mobile Robot using an Omnidirectional Camera and a Laser abstract The paper presents two different methods for mobile robot tracking and following of a fast-moving person in outdoor unstructured and possibly dynamic environment. The robot is equipped with laser range-finder and omnidirectional camera. The first method is based on visual tracking only and while it works well at slow speeds and controlled conditions, its performance quickly degrades as conditions become more difficult. The second method which uses the laser and the camera in conjunction for tracking performs well in dynamic and cluttered outdoor environments as long as the target occlusions and losses are temporary. Experimental results and analysis are presented for the second approach Marin Kobilarov, Gaurav S. Sukhatme, Jeff Hyams, Parag H. Batavia	ICRA	2
2006	A Visual Servoing Approach for Tracking Features in Urban Areas using an Autonomous Helicopter abstract The use of unmanned aerial vehicles (UAVs) in civilian and domestic applications is highly demanding, requiring a high-level of capability from the vehicles. This work addresses the design and implementation of a vision-based feature tracker for an autonomous helicopter. Using vision in the control loop allows estimating the position and velocity of a set of features with respect to the helicopter. The helicopter is then autonomously guided to track these features (in this case windows in an urban environment) in real time. The results obtained from flight trials in a real world scenario demonstrate that the algorithm for tracking features in an urban environment, used for visual servoing of an autonomous helicopter is reliable and robust Luis Mejías Alvarez, Pascual Campoy Cervera, Srikanth Saripalli, Gaurav S. Sukhatme	ICRA	4
2006	Designing Wireless Sensor Networks as a Shared Resource for Sustainable Development abstract Wireless sensor networks (WSNs) are a relatively new and rapidly developing technology; they have a wide range of applications including environmental monitoring, agriculture, and public health. Shared technology is a common usage model for technology adoption in developing countries. WSNs have great potential to be utilized as a shared resource due to their on-board processing and ad-hoc networking capabilities, however their deployment as a shared resource requires that the technical community first address several challenges. The main challenges include enabling sensor portability: (1) the frequent movement of sensors within and between deployments, and rapidly deployable systems; (2) systems that are quick and simple to deploy. We first discuss the feasibility of using sensor networks as a shared resource, and then describe our research in addressing the various technical challenges that arise in enabling such sensor portability and rapid deployment. We also outline our experiences in developing and deploying water quality monitoring wireless sensor networks in Bangladesh and California Nithya Ramanathan, Laura Balzano, Deborah Estrin, Mark H. Hansen, Thomas C. Harmon, Jenny Jay, William J. Kaiser, Gaurav S. Sukhatme	ICTD	8
2006	Engineering reliability into hybrid systems via rich design models: recent results and current directions abstract Software reliability techniques are aimed at reducing or eliminating failures in software systems. Reliability in software systems has traditionally been measured during or after system implementation. However, software engineering methodology lays stress on doing the "correct things" early on in the software development lifecycle in order to curb development and maintenance costs. In this paper, we argue that reliability of a software system should be assessed throughout the system's life span, starting with the software architecture level. Our research goal is to estimate the reliability of software systems in early design stages, which we believe involves the ability to reason about numerous uncertainties that exist in this stage, including uncertainty due to lack of execution artifacts. Our proposed approach is to develop techniques that will couple software architectural models with a suite of stochastic reliability estimation models and allow us to reason about these uncertainties. In this paper, we present our recent results using our technique for reliability estimation of software components at the level of software architecture. Another important part of this paper is the discussion of our ongoing research efforts and open research problems in this area. Somo Banerjee, Leslie Cheung, Leana Golubchik, Nenad Medvidovic, Roshanak Roshandel, Gaurav S. Sukhatme	IPDPS	6
2006	Optimum Camera Angle for Optic Flow-Based Centering Response abstract We present analytical and empirical investigations into the optimum camera angle to use for the optic flow-based centering response. This technique is commonly used to guide both ground-based and aerial robots between obstacles. A variety of camera angles have been implemented by researchers in the past, but surprisingly little mention is made of the motivation for these camera angle choices, nor has an investigation into the optimum camera angle been conducted. Our investigation shows that camera angle plays a key role in the performance of control strategies for the centering response, and both empirical and analytical investigations show the optimum camera angle to be 45 degrees when traveling between parallel obstacles Stefan Hrabar, Gaurav S. Sukhatme	IROS	2
2006	Virtual high-resolution for sensor networks abstract The resolution at which a sensor network collects data is a crucial parameter of performance since it governs the range of applications that are feasible to be developed using that network. A higher resolution, in most situations, enables more applications and improves the reliability of existing ones. In this paper we discuss a system architecture that uses controlled motion to provide virtual high-resolution in a network of cameras. Several orders of magnitude advantage in resolution may be achieved, depending on tolerable tradeoffs. We discuss several system design choices in the context of our prototype camera network implementation that realizes the proposed architecture. We also mention how some of our techniques may apply to sensors other than cameras. Real world data is collected using our prototype system and used for the evaluation of our proposed methods. Aman Kansal, William J. Kaiser, Gregory J. Pottie, Mani Srivastava 0001, Gaurav S. Sukhatme	SenSys	5
2006	Surrounding Nodes in Coordinate-Free Networks Robert Ghrist, David Lipsky, Sameera Poduri, Gaurav S. Sukhatme	WAFR	4
2006	Multirobot Simultaneous Localization and Mapping Using Manifold Representations abstract This paper describes a novel representation for two-dimensional maps, and shows how this representation may be applied to the problem of multirobot simultaneous localization and mapping. We are inspired by the notion of a manifold, which takes maps out of the two-dimensional plane and onto a surface embedded in a higher-dimensional space. The key advantage of the manifold representation is self-consistency: when closing loops, manifold maps do not suffer from the "cross over" problem exhibited in planar maps. This self-consistency, in turn, facilitates a number of important capabilities, including autonomous exploration, search, and retro-traverse. It also supports a very robust form of loop closure, in which pairs of robots act collectively to confirm or reject possible correspondence points. In this paper, we develop the basic formalism of the manifold representation, show how this may be applied to the multirobot simultaneous localization and mapping problem, and present experimental results obtained from teams of up to four robots in environments ranging in size from 400 to 900 m/sup 2/. Andrew Howard 0001, Gaurav S. Sukhatme, Maja J. Mataric	Proc. IEEE	2
2005	Networked Active Sensing of Structures Krishna Chintalapudi, John Caffrey, Ramesh Govindan, Erik A. Johnson, Bhaskar Krishnamachari, Sami F. Masri, Gaurav S. Sukhatme	DCOSS	7
2005	Coordinated Static and Mobile Sensing for Environmental Monitoring Richard Pon, Maxim A. Batalin, Victor Chen 0001, Aman Kansal, Mohammad H. Rahimi, Lisa Shirachi, Arun Somasundra, Mark H. Hansen, William J. Kaiser, Mani Srivastava 0001, Gaurav S. Sukhatme, Deborah Estrin	DCOSS	13
2005	The Analysis of an Efficient Algorithm for Robot Coverage and Exploration based on Sensor Network Deployment abstract In this paper we present the design and theoretical analysis of a novel algorithm (LRV) that efficiently solves the problems of coverage, exploration and sensor network deployment at the same time. The basic premise behind the algorithm is that the robot carries network nodes as a payload, and in the process of moving around, emplaces the nodes into the environment based on certain local criteria. In turn, the nodes emit navigation directions for the robot as it goes by. Nodes recommend directions least recently visited by the robot, hence the name LRV. We formally establish the following two properties: 1. LRV is complete on graphs, and 2. LRV is optimal on trees. We present some experimental conjectures for LRV on regular square lattice graphs and compare its performance empirically to other graph exploration algorithms. Maxim A. Batalin, Gaurav S. Sukhatme	ICRA	2
2005	Near Time-optimal Constrained Trajectory Planning on Outdoor Terrain abstract We present an outdoor terrain planner that finds near optimal trajectories under dynamic and kinematic constraints. The planner can find solutions in close to real time by relaxing some of the assumptions associated with costly rigid body simulation and complex terrain surface interactions. Our system is based on control-driven Proba bilistic Roadmaps and can efficiently find and optimize a near time-minimum trajectory. We present simulated results with artificial environments, as well as a real robot experiment using Segway Robotic Mobile Platform. Marin Kobilarov, Gaurav S. Sukhatme	ICRA	2
2005	Detection and Tracking of External Features in an Urban Environment Using an Autonomous Helicopter abstract We present the design and implementation of a real-time vision-based approach to detect and track features in a structured environment using an autonomous helicopter. Using vision as a sensor enables the helicopter to track features in an urban environment. We use vision for feature detection and a combination of vision and GPS for navigation and tracking. The vision algorithm sends high level velocity commands to the helicopter controller which is then able to command the helicopter to track them. We present results obtained from flight trials that demonstrate our algorithms for detection and tracking are applicable in real world scenarios by applying them to the task of tracking rectangular features in structured environments. Srikanth Saripalli, Gaurav S. Sukhatme, Luis Mejías Alvarez, Pascual Campoy Cervera	ICRA	2
2005	Autonomous Terrain Mapping and Classification Using Hidden Markov Models abstract This paper presents a new approach for terrain mapping and classification using mobile robots with 2D laser range finders. Our algorithm generates 3D terrain maps and classifies navigable and non-navigable regions on those maps using Hidden Markov models. The maps generated by our approach can be used for path planning, navigation, local obstacle avoidance, detection of changes in the terrain, and object recognition. We propose a map segmentation algorithm based on Markov Random Fields, which removes small errors in the classification. In order to validate our algorithms, we present experimental results using two robotic platforms. Denis F. Wolf, Gaurav S. Sukhatme, Dieter Fox, Wolfram Burgard	ICRA	2
2005	Robomote: enabling mobility in sensor networks abstract Severe energy limitations, and a paucity of computation pose a set of difficult design challenges for sensor networks. Recent progress in two seemingly disparate research areas namely, distributed robotics and low power embedded systems has led to the creation of mobile (or robotic) sensor networks. Autonomous node mobility brings with it its own challenges, but also alleviates some of the traditional problems associated with static sensor networks. We illustrate this by presenting the design of the robomote, a robot platform that functions as a single mobile node in a mobile sensor network. We briefly describe two case studies where the robomote has been used for table top experiments with a mobile sensor network. Karthik Dantu, Mohammad H. Rahimi, Hardik Shah, Sandeep Babel, Amit Dhariwal, Gaurav S. Sukhatme	IPSN	6
2005	Networked infomechanical systems: a mobile embedded networked sensor platform abstract Networked infomechanical systems (NIMS) introduces a new actuation capability for embedded networked sensing. By exploiting a constrained actuation method based on rapidly deployable infrastructure, NIMS suspends a network of wireless mobile and fixed sensor nodes in three-dimensional space. This permits run-time adaptation with variable sensing location, perspective, and even sensor type. Discoveries in NIMS environmental investigations have raised requirements for 1) new embedded platforms integrating many diverse sensors with actuators, and 2) advances for in-network sensor data processing. This is addressed with a new and generally applicable processor-preprocessor architecture described in this paper. Also this paper describes the successful integration of R, a powerful statistical computing environment, into the embedded NIMS node platform. Richard Pon, Maxim A. Batalin, Jason Gordon, Aman Kansal, Mohammad H. Rahimi, Lisa Shirachi, Mark H. Hansen, William J. Kaiser, Mani Srivastava 0001, Gaurav S. Sukhatme, Deborah Estrin	IPSN	12
2005	Task allocation for event-aware spatiotemporal sampling of environmental variables abstract Monitoring of environmental phenomena with embedded networked sensing confronts the challenges of both unpredictable variability in the spatial distribution of phenomena coupled with the demands for a high spatial sampling rate in three dimensions. For example, low distortion mapping of critical solar radiation properties in forest environments may require two-dimensional spatial sampling rates of greater than 10 samples/m/sup 2/ over transects exceeding 1000 m/sup 2/. Clearly, adequate sampling coverage of such transect requires an impractically large number of sensing nodes. A new approach, networked infomechanical system (NIMS), has been introduced to combine autonomous-articulated and static sensor nodes enabling sufficient spatiotemporal sampling density over large transects to meet a general set of environmental mapping demands. This paper describes our work on the critical parts of NIMS, the task allocation module. We present our methodologies and the two basic greedy task allocation policies - based on time of the task arrival (time policy) and distance from the robot to the task (distance policy). We present results from NIMS deployed in a forest reserve and from a lab testbed. The results show that both policies are adequate for the task of spatiotemporal sampling, but also complement each other. Finally, we suggest the future direction of research that would both help us better quantify the performance of our system and create more complex policies. Maxim A. Batalin, Gaurav S. Sukhatme, Richard Pon, Jason Gordon, Mohammad H. Rahimi, William J. Kaiser, Gregory J. Pottie, Deborah Estrin	IROS	2
2005	Combined optic-flow and stereo-based navigation of urban canyons for a UAV abstract We present a novel vision-based technique for navigating an unmanned aerial vehicle (UAV) through urban canyons. Our technique relies on both optic flow and stereo vision information. We show that the combination of stereo and optic flow (stereo flow) is more effective at navigating urban canyons than either technique alone. Optic flow from a pair of sideways looking cameras is used to stay centered in a canyon and initiate turns at junctions, while stereo vision from a forward facing stereo head is used to avoid obstacles to the front. The technique was tested in full on an autonomous tractor at CSIRO and in part on the USC autonomous helicopter. Experimental results are presented from these two robotic platforms operating in outdoor environments. We show that the autonomous tractor can navigate urban canyons using stereo-flow, and that the autonomous helicopter can turn away from obstacles to the side using optic flow. In addition, preliminary results show that a single pair of forward facing fisheye cameras can be used for both stereo and optic flow. The center portions of the fisheye images are used for stereo, while flow is measured in the periphery of the images. Stefan Hrabar, Gaurav S. Sukhatme, Peter I. Corke, Kane Usher, Jonathan Roberts 0001	IROS	2
2005	Adaptive sampling for environmental field estimation using robotic sensors abstract Monitoring environmental phenomena by distributed sensor sampling confronts the challenge of unpredictable variability in the spatial distribution of phenomena often coupled with demands for a high spatial sampling rate. The introduction of actuation-enabled robotics sensors permits a system to optimize the sampling distribution through runtime adaptation. However, such systems must efficiently dispense sampling points or otherwise suffer from poor temporal response. In this paper, we propose and characterize an active modeling system. In our approach, as the robotic sensor acquires measurement samples of the environment, it builds a model of the phenomenon. Our algorithm is based on an incremental optimization process where the robot supports a continuous, iterative process of 1) collecting samples with maximal coverage in the design space; 2) building the environmental model; 3) predicting sampling point locations that contribute the greatest certainty regarding the phenomenon; and 4) sampling the environment based on a combined measure of information gain and navigation and sampling cost. This can provide significant reductions in the magnitude of field estimation error with a modest navigational trajectory time. We evaluate our algorithm through a simulation, using a combination of static and mobile sensors sampling light illumination field. Mohammad H. Rahimi, Mark H. Hansen, William J. Kaiser, Gaurav S. Sukhatme, Deborah Estrin	IROS	4
2005	Towards geometric 3D mapping of outdoor environments using mobile robots abstract This paper presents an approach to generating compact 3D maps of urban environments using mobile robots and laser range finders. Our algorithm extracts planar information from 3D point cloud maps. The planar representation is very efficient for representing building structures in urban environments when a high level of detail is not required. We also present preliminary results on 3D geometric mapping with incomplete data. Based on previously known models and incomplete data, our system is able to estimate parts of buildings which have never been seen before. As validation, we present experimental results using a Segway RMP vehicle in two environments, both approximately the size of a city block. Denis F. Wolf, Andrew Howard 0001, Gaurav S. Sukhatme	IROS	3
2005	Bias Reduction and Filter Convergence for Long Range Stereo Gabe Sibley, Larry H. Matthies, Gaurav S. Sukhatme	ISRR	3
2004	Using a Sensor Network for Distributed Multi-robot Task Allocation abstract We present a multi field distributed in-network task allocation (DINTA-MF) algorithm for online multi-robot task allocation (OMRTA) where tasks are allocated explicitly to robots by a pre-deployed, static sensor network. The idea of DINTA-MF is to compute several assignment fields in the sensor network and then distributively assign fields to different robots. Experimental results with a simulated alarm scenario show that our approach is able to compute solutions to the OMRTA problem in a distributed fashion and arguably in an optimal way. We compared DINTA-MF with a simpler implementation (DINTA), which uses one assignment field. The data show that DINTA-MF outperforms DINTA as the number of robots increases. Maxim A. Batalin, Gaurav S. Sukhatme	ICRA	2
2004	Mobile Robot Navigation Using a Sensor Network abstract We describe an algorithm for robot navigation using a sensor network embedded in the environment. Sensor nodes act as signposts for the robot to follow, thus obviating the need for a map or localization on the part of the robot. Navigation directions are computed within the network (not on the robot) using value iteration. Using small low-power radios, the robot communicates with nodes in the network locally, and makes navigation decisions based on which node it is near. An algorithm based on processing of radio signal strength data was developed so the robot could successfully decide which node neighborhood it belonged to. Extensive experiments with a robot and a sensor network confirm the validity of the approach. Maxim A. Batalin, Gaurav S. Sukhatme, Myron Hattig	ICRA	2
2004	Autonomous Deployment and Repair of a Sensor Network using an Unmanned Aerial Vehicle abstract We describe a sensor network deployment method using autonomous flying robots. Such networks are suitable for tasks such as large-scale environmental monitoring or for command and control in emergency situations. We describe in detail the algorithms used for deployment and for measuring network connectivity and provide experimental data we collected from field trials. A particular focus is on determining gaps in connectivity of the deployed network and generating a plan for a second, repair, pass to complete the connectivity. This project is the result of a collaboration between three robotics labs (CSIRO, USC, and Dartmouth.). Peter I. Corke, Stefan Hrabar, Ronald A. Peterson, Daniela Rus, Srikanth Saripalli, Gaurav S. Sukhatme	ICRA	6
2004	Bacterium-inspired Robots for Environmental Monitoring abstract Locating gradient sources and tracking them over time has important applications to environmental monitoring and studies of the ecosystem. We present an approach, inspired by bacterial chemotaxis, for robots to navigate to sources using gradient measurements and a simple actuation strategy (biasing a random walk). Extensive simulations show the efficacy of the approach in varied conditions including multiple sources, dissipative sources, and noisy sensors and actuators. We also show how such an approach could be used for boundary finding. We validate our approach by testing it on a small robot (the robomote) in a phototaxis experiment. A comparison of our approach with gradient descent shows that while gradient descent is faster, our approach is better suited for boundary coverage, and performs better in the presence of multiple and dissipative sources. Amit Dhariwal, Gaurav S. Sukhatme, Aristides A. G. Requicha	ICRA	2
2004	A Generalized Region-based Approach for Multi-target Tracking in Outdoor Environments abstract We propose a generalized region-based approach to multi-target tracking, which is applicable to structured and unstructured environments. In this approach each robot constructs virtual regions based on the latest tracking information from other robots. Without pre-partitioned region information, each robot independently estimates the most urgent region that needs to be visited. The idea is for robots to coarsely estimate where the targets are present, and to navigate there. A multi-robot system to track moving objects outdoors has been designed using this approach in order to validate the idea. The performance of the individual motion tracker; and the cooperative tracking behaviors is evaluated through experiments with different robot bases (a helicopter, a Segway RMP, and a Pioneer) and in simulation. Experimental results indicate that robots are able to distribute themselves appropriately in response to target movement. Boyoon Jung, Gaurav S. Sukhatme	ICRA	2
2004	Constrained Coverage for Mobile Sensor Networks abstract We consider the problem of self-deployment of a mobile sensor network. We are interested in a deployment strategy that maximizes the area coverage of the network with the constraint that each of the nodes has at least K neighbors, where K is a user-specified parameter. We propose an algorithm based on artificial potential fields which is distributed, scalable and does not require a prior map of the environment. Simulations establish that the resulting networks have the required degree with a high probability, are well connected and achieve good coverage. We present analytical results for the coverage achievable by uniform random and symmetrically tiled network configurations and use these to evaluate the performance of our algorithm. Sameera Poduri, Gaurav S. Sukhatme	ICRA	2
2004	Adaptive Sampling for Environmental Robotics abstract The capabilities and distributed nature of networked sensors are uniquely suited to the characterization of distributed phenomena in the natural environment. However, environmental characterization by fixed distributed sensors encounters challenges in complex environments. In this paper we describe Networked Infomechanical Systems (NIMS), a new distributed, robotic sensor methodology developed for applications including characterization of environmental structure and phenomena. NIMS exploits deployed infrastructure that provides the benefits of precise motion, aerial suspension, and low energy sustainable operations in complex environments. NIMS nodes may explore a three-dimensional environment and enable the deployment of sensor nodes at diverse locations and viewing perspectives. NIMS characterization of phenomena in a three dimensional space must now consider the selection of sensor sampling points in both time and space. Thus, we introduce a new approach of mobile node adaptive sampling with the objective of minimizing error between the actual and reconstructed spatiotemporal behavior of environmental variables while minimizing required motion. In this approach, the NIMS node first explores as an agent, gathering a statistical description of phenomena using a nested stratified random sampling approach. By iteratively increasing sampling resolution, guided adaptively by the measurement results themselves, this NIMS sampling enables reconstruction of phenomena with a systematic method for balancing accuracy with sampling resource cost in time and motion. This adaptive sampling method is described analytically and also tested with simulated environmental data. Experimental evaluations of adaptive sampling algorithms have also been completed. Specifically, NIMS experimental systems have been developed for monitoring of spatiotemporal variation of atmospheric climate phenomena. A NIMS system has been deployed at a field biology station to map phenomena in a 50m width and 50m span transect in a forest environment. In addition, deployments have occurred in testbed environments allowing additional detailed characterization of sampling algorithms. Environmental variable mapping of temperature, humidity, and solar illumination have been acquired and used to evaluate the adaptive sampling methods reported here. These new methods have been shown to provide a significant advance for efficient mapping of spatially distributed phenomena by NIMS environmental robotics. Mohammad H. Rahimi, Richard Pon, William J. Kaiser, Gaurav S. Sukhatme, Deborah Estrin, Mani Srivastava 0001	ICRA	4
2004	A Multi-robot Approach to Stealthy Navigation in the Presence of an Observer abstract We propose a simple, reactive method for multiple robots carrying out sequential low-visibility navigation in the presence of an observer. Initially, the robots have no map of the environment but know the locations of the observer and goal. They generate an occupancy grid representation of the environment which is modeled using potential fields with embedded task information. These fields are combined and navigation waypoints extracted. Each robot carries out its traverse independently and shares its experience with its successor. The experience information consists of the occupancy grid and a filtered version of the traveled path used to assist the subsequent robot to traverse a lower visibility path. This produces a robust and reactive solution for stealthy navigation since there is no global path planning and the robots are not committed to any particular path. Experiments in simulation and real outdoor environments substantiate the approach and demonstrate the benefits of sharing information in reducing cumulative visibility. The experiments also demonstrate the algorithm's versatility in taking advantage of an environment that changes between robot traverses. Ashley Tews, Gaurav S. Sukhatme, Maja J. Mataric	ICRA	2
2004	Online Simultaneous Localization and Mapping in Dynamic Environments abstract We propose an on-line algorithm for simultaneous localization and mapping of dynamic environments. Our algorithm is capable of differentiating static and dynamic parts of the environment and representing them appropriately on the map. Our approach is based on maintaining two occupancy grids. One grid models the static parts of the environment, and the other models the dynamic parts of the environment. The union of the two provides a complete description of the environment over time. We also maintain a third map containing information about static landmarks detected in the environment. These landmarks provide the robot with localization. Results in simulation and with physical robots show the efficiency of our approach and show how the differentiation of dynamic and static entities in the environment and SLAM can be mutually beneficial. Denis F. Wolf, Gaurav S. Sukhatme	ICRA	2
2004	Ad-Hoc Localization Using Ranging and Sectoring abstract Ad-hoc localization systems enable nodes in a sensor network to fix their positions in a global coordinate system using a relatively small number of anchor nodes that know their position through external means (e.g., GPS). Because location information provides context to sensed data, such systems are a critical component of many sensor networks and have therefore received a fair amount of recent attention in the sensor networks literature. The efficacy of these systems is a function of the density of deployment and of anchor nodes, as well as the error in distance estimation (ranging) between nodes. In this paper, we examine how these factors impact the performance of the system. This examination lays the groundwork for the main question we consider in this paper: Can the ability to estimate bearing to neighboring nodes greatly increase the performance of ad-hoc localization systems? We discuss the design of ad-hoc localization systems that use range together with either bearing or imprecise bearing (such as sectoring) information, and evaluate these systems using analysis and simulation. Krishna Chintalapudi, Ramesh Govindan, Gaurav S. Sukhatme, Amit Dhariwal	INFOCOM	3
2004	Towards 3D mapping in large urban environments abstract This paper describes work-in-progress aimed at generating dense 3D maps of urban environments using laser range data acquired from a moving platform. These maps display both fine-scale detail (resolving features only a few centimeters across) and large-scale consistency (typical maps are approximately 0.5 km on a side). In this paper, we sketch a basic 3D mapping algorithm (paying particular attention to practical engineering details) and present preliminary results acquired on the USC University Park campus using a Segway RMP vehicle. Andrew Howard 0001, Denis F. Wolf, Gaurav S. Sukhatme	IROS	3
2004	A comparison of two camera configurations for optic-flow based navigation of a UAV through urban canyons abstract We present a comparison of two camera configurations for avoiding obstacles in 3D-space using optic flow. The two configurations were developed for use on an autonomous helicopter, with the aim of enabling it to fly in environments with tall obstacles (e.g. urban canyons). The comparison is made based on real data captured from two sideways-looking cameras and an omnidirectional camera mounted onboard an autonomous helicopter. Optic flow information from the images is used to determine the relative distance to obstacles on each side of the helicopter. We show that on average, both camera configurations are equally effective and that they can be used to tell which of the canyon walls is closer with an accuracy of 74%. It is noted that each configuration is however more effective under certain conditions, and so a suitable hybrid approach is suggested. We also show that there is a linear relationship between the optic flow ratios and the position of the helicopter with respect to the center of the canyon. We use this relationship to develop a proportional control strategy for flying the helicopter along the Voronoi line between buildings. Stefan Hrabar, Gaurav S. Sukhatme	IROS	2
2004	Detecting anomalous human interactions using laser range-finders abstract We present a laser range-finder-based system for tracking people in an outdoor environment and detecting interactions between them. The system does not use identities of people for tracking. Observed tracks are automatically segmented into individual activities using an entropy-based measure (Jensen-Shannon divergence (Lin, J, 1991)). Two people situated close to each other throughout the duration of an activity represents an interaction. The observed activities are combined using a hierarchical clustering algorithm to generate a representative set. The frequency of occurrence of these activities is modeled by a Poisson distribution. During the monitoring phase, this model is used to compute the probability of observing the detected activities and interactions; an anomaly is flagged if this probability falls below a threshold. Experimental results from an outdoor courtyard environment are described where the system indicates anomalies when there is a sudden increase in the number of people in the environment or in the number of interactions. This detection occurs without giving the system any a priori concepts of space occupancy. Anand Panangadan, Maja J. Mataric, Gaurav S. Sukhatme	IROS	3
2004	Avoiding detection in a dynamic environment abstract Remaining elusive while navigating to a goal in a dynamic environment containing an observer requires taking advantage of opportunistic cover as it occurs. A reactive navigation approach is needed that recognizes the utility of environment features in offering protective cover. We present an approach that allows stealthy traverses in unknown environments containing dynamic objects. It is a frontier-based method that allows a robot to follow in the obscuring shadow of objects despite their dynamics, and take advantage of more opportunistic cover if it becomes available. An analysis of our approach in off-line modeling and experiments conducted in simulation and outdoor environments demonstrate its effectiveness in achieving high quality solutions for stealthy navigation. Ashley Tews, Maja J. Mataric, Gaurav S. Sukhatme	IROS	3
2004	Adaptive sampling for marine microorganism monitoring abstract We describe the design and construction of an underwater sensor actuator network to detect extreme temperature gradients. We are motivated by the fact that regions of sharp temperature change (thermoclines) are a breeding ground for certain marine microorganisms. We present a distributed algorithm using local communication based on binary search to find a thermocline by using a mobile sensor network. Simulations and experiments using a mote test bed demonstrate the validity of this approach. We also discuss the improvement in energy efficiency using a submarine robot as a data mule. Comparisons between experimental data with and without the data mule show that there are considerable energy savings in the sensor network due to the data mule. Bin Zhang 0010, Gaurav S. Sukhatme, Aristides A. G. Requicha	IROS	2
2004	Call and response: experiments in sampling the environment abstract Monitoring of environmental phenomena with embedded networked sensing confronts the challenges of both unpredictable variability in the spatial distribution of phenomena, coupled with demands for a high spatial sampling rate in three dimensions. For example, low distortion mapping of critical solar radiation properties in forest environments may require two-dimensional spatial sampling rates of greater than 10 samples/m2 over transects exceeding 1000 m2. Clearly, adequate sampling coverage of such a transect requires an impractically large number of sensing nodes. This paper describes a new approach where the deployment of a combination of autonomous-articulated and static sensor nodes enables sufficient spatiotemporal sampling densityo ver large transects to meet a general set of environmental mapping demands.To achieve this we have developed an embedded networked sensor architecture that merges sensing and articulation with adaptive algorithms that are responsive to both variabilityin environmental phenomena discovered bythe mobile sensors and to discrete events discovered byst atic sensors. We begin byde scribing the class of important driving applications, the statistical foundations for this new approach, and task allocation. We then describe our experimental implementation of adaptive, event aware, exploration algorithms, which exploit our wireless, articulated sensors operating with deterministic motion over large areas. Results of experimental measurements and the relationship among sampling methods, event arrival rate, and sampling performance are presented. Maxim A. Batalin, Mohammad H. Rahimi, Aman Kansal, Gaurav S. Sukhatme, William J. Kaiser, Mark H. Hansen, Gregory J. Pottie, Mani Srivastava 0001, Deborah Estrin	SenSys	6
2004	A sensor-actuator network for damage detection in civil structures abstract Structural health monitoring (SHM) is a well-established multi-disciplinary research field. The goal of SHM is to develop technologies and techniques to automatically detect, localize, and classify damages in large structures (ships, bridges, aircraft and buildings). The state of the art in SHM relies on collecting response of these structures to ambient phenomena such as wind, passing vehicles or earthquakes at various points in the structure (either via manual inspections or expensive wired data acquisition systems) to be analyzed centrally. In our demonstration we will show a proof of concept working model of an automated distributed damage detection system using a sensor-actuator network. Krishna Chintalapudi, Karthik Dantu, Sandeep Babel, Ramesh Govindan, Gaurav S. Sukhatme, John Caffrey	SenSys	5
2003	Efficient exploration without localization abstract We study the problem of exploring an unknown environment using a single robot. The environment is large enough (and possibly dynamic) that constant motion by the robot is needed to cover the environment. We term this the dynamic coverage problem. We present an efficient minimalist algorithm which assumes that global information is not available to the robot (neither a map, nor GPS). Our algorithm uses markers which the robot drops off as signposts to aid exploration. We conjecture that our algorithm has a cover time better than O(n log n), where the n markers that are deployed form the vertices of a regular graph. We provide experimental evidence in support of this conjecture. We show empirically that the performance of our algorithm on graphs is similar to its performance in simulation. Maxim A. Batalin, Gaurav S. Sukhatme	ICRA	2
2003	Multi-robot task-allocation through vacancy chains abstract This paper presents an algorithm for task allocation in groups of homogeneous robots. The algorithm is based on vacancy chains, a resource distribution strategy common in human and animal societies. We define a class of task-allocation problems for which the vacancy chain algorithm is suitable and demonstrate how reinforcement learning can be used to make vacancy chains emerge in a group of behavior-based robots. Experiments in simulation show that the vacancy chain algorithm consistently outperforms random and static task allocation algorithms when individual robots are prone to distractions or breakdowns, or when task priorities change. Torbjørn S. Dahl, Maja J. Mataric, Gaurav S. Sukhatme	ICRA	3
2003	Putting the 'I' in 'team': an ego-centric approach to cooperative localization abstract This paper describes a cooperative method for relative localization of mobile robot teams; that is, it describes a method whereby every robot in the team can estimate the pose of every other robot, relative to itself. This robot does not require the use of GPS, landmarks, or maps of any kind; instead, robots make direct measurement of the relative pose of nearby robots, and broadcast this information to the team as a whole. Each robot processes this information independently to generate ego-centric estimate for the pose of other robots. Our method uses Bayesian formalism with a particle filter implementation, and is, as a consequence, very robust. It is also completely distributed, yet requires relatively little communication between robots. This paper describes the basic ego-centric formalism, sketches the implementation, and presents experimental results obtained using a team of four mobile robots. Andrew Howard 0001, Maja J. Mataric, Gaurav S. Sukhatme	ICRA	3
2003	Onmidirectional vision for an autonomous helicopter abstract We present the design and implementation of an omnidirectional vision system used for sideways-looking sensing on an autonomous helicopter. To demonstrate the capabilities of the system, a visual servoing task was designed which required the helicopter to locate and move towards the centroid of a number of visual targets. Results are presented showing that the task was successfully completed by a Pioneer ground robot equipped with the same omnidirectional vision system, and preliminary test flight results show that the system can generate appropriate control commands for the helicopter. Stefan Hrabar, Gaurav S. Sukhatme	ICRA	2
2003	Studying the Feasibility of Energy Harvesting in a Mobile Sensor Network abstract We study the feasibility of extending the lifetime of a wireless sensor network by exploiting mobility. In our system, a small percentage of network nodes are autonomously mobile, allowing them to move in search of energy, recharge, and delivery energy to immobile, energy-depleted nodes. We term this approach energy harvesting. We characterize the problem of uneven energy consumption, suggest energy harvesting as a possible solution, and provide a simple analytical framework to evaluate energy consumption and our scheme. Data from initial feasibility experiments using energy harvesting show promising results. Mohammad H. Rahimi, Hardik Shah, Gaurav S. Sukhatme, John S. Heidemann, Deborah Estrin	ICRA	3
2003	A scalable approach to human-robot interaction abstract Much of the current research in human-robot interaction is concerned with single systems and single or few users. These systems and their interfaces are generally tightly-coupled and well-defined. For large-scale human-robot applications, the systems may be unknown prior to designing the interface for potential human interaction. This presents a difficult goal for allowing multiple users to interact with many possibly unknown systems. In this paper, we present an interaction infrastructure aligned with providing this interface. It operates in two phases that accommodate both many-to-many interaction and generalized, one-to-one interaction between users and robotic systems. Our previous research has demonstrated the infrastructure to scale to a large number of users and several systems in simulation. The experiments in this paper substantiate these results in a smaller-scale real robotic environment. Ashley Tews, Maja J. Mataric, Gaurav S. Sukhatme	ICRA	3
2003	Sensor network-based multi-robot task allocation abstract We present DINTA, distributed in-network task allocation - a novel paradigm for multi-robot task allocation (MRTA) where tasks are allocated implicitly to robots by a pre-deployed, static sensor network. Experimental results with a simulated alarm scenario show that our approach is able to compute solutions to the MRTA problem in a distributed fashion. We compared our approach to a strategy where robots use the deployed sensor network for efficient exploration. The data show that our approach outperforms such an exploration-only algorithm. The data also provide evidence that the proposed algorithm is more stable than the exploration-only algorithm. Maxim A. Batalin, Gaurav S. Sukhatme	IROS	2
2003	Towards stealthy behaviors abstract This paper describes a behavior-based approach to stealth. The specific problem we consider is that of making stealthy traverses, i.e., transiting from one point to another while remaining hidden from an observer. Since we assume that the robot has no a priori model of the environment, our stealthy traverse behavior makes opportunistic use of terrain features to hide from the observer. This behavior has been evaluated in both real and simulated experiments, comparing it against a regular goal-seeking/obstacle-avoidance behavior. These experiments show a clear improvement in the stealthiness of the robot. Emil Birgersson, Andrew Howard 0001, Gaurav S. Sukhatme	IROS	3
2003	A tale of two helicopters abstract This paper discusses similarities and differences in autonomous helicopters developed at USC and CSIRO. The most significant differences are in the accuracy and sample rate of the sensor systems used for control. The USC vehicle, like a number of others, makes use of a sensor suite that costs an order of magnitude more than the vehicle. The CSIRO system, by contrast, utilizes low-cost inertial, magnetic, vision and GPS to achieve the same ends. We describe the architecture of both autonomous helicopters, discuss the design issues and present comparative results. Srikanth Saripalli, Jonathan Roberts 0001, Peter I. Corke, Gregg D. Buskey, Gaurav S. Sukhatme	IROS	5
2003	Sensor network as a distributed manager for multi-robot task allocation Maxim A. Batalin, Gaurav S. Sukhatme	SenSys	2
2003	Contour detection using actuated sensor networks abstract No abstract available. Karthik Dantu, Gaurav S. Sukhatme	SenSys	2
2003	Visually guided landing of an unmanned aerial vehicle abstract We present the design and implementation of a real-time, vision-based landing algorithm for an autonomous helicopter. The landing algorithm is integrated with algorithms for visual acquisition of the target (a helipad) and navigation to the target, from an arbitrary initial position and orientation. We use vision for precise target detection and recognition, and a combination of vision and Global Positioning System for navigation. The helicopter updates its landing target parameters based on vision and uses an onboard behavior-based controller to follow a path to the landing site. We present significant results from flight trials in the field which demonstrate that our detection, recognition, and control algorithms are accurate, robust, and repeatable. Srikanth Saripalli, James F. Montgomery, Gaurav S. Sukhatme	IEEE Trans. Robotics Autom.	3
2002	Exploiting Physical Dynamics for Concurrent Control of a Mobile Robot abstract Conventionally, mobile robots are controlled through an action selection mechanism (ASM) that chooses among multiple proposed actions. This choice can be made in a variety of ways, and ASMs have been developed that demonstrate many of them, from strict priority schemes to voting systems. We take a different approach, which we call concurrent control. Abandoning explicit action selection, we rely instead on the physical dynamics of the robot's actuators to achieve robust control. We claim that with many noisy controllers and no arbitration among commands we can elicit stable predictable behavior from a mobile robot. Specifically, we concern ourselves with the problem of driving a single planar mobile robot with multiple controlling agents that are concurrently sending commands directly to the robot's wheel motors. We state analytically conditions necessary for our concurrent control approach to be viable, and verify empirically its general effectiveness and robustness to error through experiments with a physical robot. Brian P. Gerkey, Maja J. Mataric, Gaurav S. Sukhatme	ICRA	3
2002	Controlling Hopping Height of a Pneumatic Monopod abstract We describe a model-based height controller for a hopping robot with a pneumatically powered leg. The controller explicitly models variation in the leg angle and height. Using an explicit model of the physics of the pneumatic spring and some symmetry assumptions, we derive the desired leg-length setting to regulate apex hopping height using a PD controller. Simulation experiments of hopping in the sagittal plane show reasonable height regulation. For low-speed running, we take advantage of the small variations in the leg angle about the vertical, and demonstrate that the original symmetry assumptions may be relaxed by restricting the leg angle to /spl pi//2. Simulations show that the restricted model outperforms the original model by a small but significant amount. Kale Harbick, Gaurav S. Sukhatme	ICRA	2
2002	Multi-Robot Task Allocation in the Light of Uncertainty abstract We describe an empirical study that sought general guidelines for task allocation strategies in multi-robot systems. We identify four distinct task allocation strategies, and demonstrate them in two versions of the multi-robot emergency handling task. We describe an experimental setup to compare results obtained from a simulated grid world to the results from real world experiments. Data resulting from eight hours of real mobile robot experiments are compared to the trend identified in simulation. The data from the simulations show that there is no single strategy that produces best performance in all cases, and that the best task allocation strategy changes as a function of the noise in the system. This result is significant, and shows the need for further investigation of task allocation strategies. Esben Hallundbæk Østergaard, Maja J. Mataric, Gaurav S. Sukhatme	ICRA	3
2002	Vision-Based Autonomous Landing of an Unmanned Aerial Vehicle abstract We present the design and implementation of a real-time, vision-based landing algorithm for an autonomous helicopter. The helicopter is required to navigate from an initial position to a final position in a partially known environment based on GPS and vision, locate a landing target (a helipad of a known shape) and land on it. We use vision for precise target detection and recognition. The helicopter updates its landing target parameters based on vision and uses an on-board behavior-based controller to follow a path to the landing site. We present results from flight trials in the field which demonstrate that our detection, recognition and control algorithms are accurate and repeatable. Srikanth Saripalli, James F. Montgomery, Gaurav S. Sukhatme	ICRA	3
2002	Robomote: A Tiny Mobile Robot Platform for Large-Scale Ad-Hoc Sensor Networks abstract This paper introduces Robomote, a robotic solution developed to explore problems in large-scale distributed robotics and sensor networks. The design explicitly aims at enabling research in sensor networking, adhoc networking, massively distributed robotics, and extended longevity. The platform must meet many demanding criteria not limited to but including: miniature size, low power, low cost, simple fabrication, and a sensor/actuator suite that facilitates navigation and localization. We argue that a robot test bed such as Robomote is necessary for practical research with large networks of mobile robots. Further, we present a preliminary analysis of Robomotes' success to this end. Gabe Sibley, Mohammad H. Rahimi, Gaurav S. Sukhatme	ICRA	3
2002	Staying Alive: A Docking Station for Autonomous Robot Recharging abstract Autonomous mobile robots are constrained in their long-term functionality due to a limited on-board power supply. Typically, rechargeable batteries are utilized that may only provide a few hours of peak usage before recharging is necessary. Recharging requires a robot to be taken offline, and attached to a battery charger via human intervention. This is unacceptable in environments where long-term autonomous capabilities are necessary. We present a method to provide long-term autonomy by implementing autonomous recharging. A recharging station design is presented, consisting of a stationary docking station and a docking mechanism mounted to a Pioneer 2DX robot. The docking station and robot docking mechanism are designed to work together, providing a mechanical and electrical connection between the charging system and the robot. Algorithms are implemented to monitor the battery voltage and control the docking procedure, as well as account for any errors that may occur. Initial experiments that demonstrate the validity of the approach and design are presented. Milo C. Silverman, Dan Nies, Boyoon Jung, Gaurav S. Sukhatme	ICRA	4
2002	Exploiting Task Regularities to Transform Between Reference Frames in Robot Teams abstract We describe a team of robots that uses a trail of landmarks to navigate between places of interest. The landmarks are not physical; they are waypoint coordinates generated online by each robot and shared with teammates over the network. Waypoints are specified with reference to features in the world that are relevant to the team's task and common to all robots. Using such task-level features as landmarks avoids the need to sense and name physical landmarks. Using these common landmarks, each robot can transform waypoint coordinates into its local reference frame, avoiding the cost of maintaining a fixed global coordinate system. The algorithm is tested in an experiment in which a team of 4 autonomous mobile robots run in our office building for more than 3 hours, travelling a total of 8.2 km (5.1 miles). Despite significant divergence of their local coordinate systems, they are able to share waypoints, forming and following a common trail between two fixed locations. Richard Vaughan 0001, Kasper Støy, Gaurav S. Sukhatme, Maja J. Mataric	ICRA	3
2002	Adaptive spatio-temporal organization in groups of robots abstract This paper presents experiments, in simulation, with a group of robots that improve their performance on a straightforward transportation task by using reinforcement learning to associate input states with a set of abstract behaviors. We show that the improvement in performance is a result of the group adapting its spatio-temporal organization to the given environment. Spatio-temporal adaptation is a general form of adaptation in that it can improve performance over a range of different tasks and environments. Hence it increases the general applicability and autonomy of robotic systems. Lastly, we present two communication strategies that improve this ability to adapt by generally improving learning rates for cooperative robots in highly dynamic domains. Torbjørn S. Dahl, Maja J. Mataric, Gaurav S. Sukhatme	IROS	3
2002	Localization for mobile robot teams using maximum likelihood estimation abstract This paper describes a method for localizing the members of a mobile robot team, using only the robots themselves as landmarks, that is, we describe a method whereby each robot can determine the relative range, bearing and orientation of every other robot in the team, without the use of GPS, external landmarks, or instrumentation of the environment. Our method assumes that each robot is able to measure the relative pose of nearby robots, together with changes in its own pose. Using a combination of maximum likelihood estimation and numerical optimization, we can subsequently infer the relative pose of every robot in the team. This paper describes the basic formalism, its practical implementation, and presents experimental results obtained using a team of four mobile robots. Andrew Howard 0001, Maja J. Matark, Gaurav S. Sukhatme	IROS	3
2002	An incremental deployment algorithm for mobile robot teams abstract This paper describes an algorithm for deploying the members of a mobile robot team into an unknown environment. The algorithm deploys robots one-at-a-time, with each robot making use of information gathered by the previous robots to determine the next deployment location. The deployment pattern is designed to maximize the area covered by the robots' sensors, while simultaneously ensuring that the robots maintain line-of-sight contact with one another This paper describes the basic algorithm and presents results obtained from a series of experiments conducted using both real and simulated robots. Andrew Howard 0001, Maja J. Mataric, Gaurav S. Sukhatme	IROS	3
2002	A region-based approach for cooperative multi-target tracking in a structured environment abstract This paper addresses the problem of tracking multiple targets using a network of communicating robots and stationary sensors. We introduce a region-based approach which controls robot deployment at two levels. A coarse deployment controller distributes robots across regions using a topological map and density estimates, and a target-following controller attempts to maximize the number of tracked targets within a region. A behavior-based system is presented implementing the region-based approach. Intensive simulations were performed to investigate the correlation between our approach and the degree of occlusion in the environment. The region-based approach shows better performance than a 'naive' local-following strategy when the environment has significant occlusion. We performed real-robot experiments to validate the system. These experiments open up a new line of research, which suggests that an optimal ratio of robots to stationary sensors may exist for a given environment with certain occlusion characteristics. Boyoon Jung, Gaurav S. Sukhatme	IROS	2
2002	An implicit-based haptic rendering technique abstract We present a novel haptic rendering technique. Building on previous work, we propose a haptic model based on a volumetric description of the geometry of an object. Unlike previous volumetric approaches, we also find a virtual contact point on the surface in order to derive a penalty force that is consistent with the real geometry of the object, without introducing force discontinuity. We also demonstrate that other surface properties such as friction and texture can be added elegantly. The resulting technique is fast (a constant 1000 Hz refresh rate) and can handle large geometry models on low-end computers. Laehyun Kim, Anna Kyrikou, Gaurav S. Sukhatme, Mathieu Desbrun	IROS	3
2002	Haptic control of a mobile robot: a user study abstract We address the problem of teleoperating a mobile robot using shared autonomy: an on-board controller performs obstacle avoidance while the operator uses the manipulandum of a haptic probe to designate the desired speed and rate of turn. Sensors on the robot are used to measure obstacle range information. We describe a strategy to convert such range information into forces, which are reflected to the operator's hand, via the haptic probe. This haptic information provides feedback to the operator in addition to imagery from a front-facing camera mounted on the mobile robot. Extensive experiments with a user population show that the added haptic feedback significantly improves operator performance in several ways (reduced collisions, increased minimum distance between the robot and obstacles) without a significant increase in navigation time. Sangyoon Lee 0004, Gaurav S. Sukhatme, Gerard Jounghyun Kim, Chan-Mo Park	IROS	2
2002	A testbed for Mars precision landing experiments by emulating spacecraft dynamics on a model helicopter abstract We propose the use of a model helicopter to emulate the landing dynamics of a spacecraft. Our controller accepts thruster inputs (like those on a spacecraft) and converts them into appropriate helicopter stick controls such that the resulting trajectory of the helicopter is close to the trajectory that would have been achieved by simply providing the same thruster inputs to a spacecraft. The approach relies on simplified models of the spacecraft and helicopter dynamics. Initial results in simulation indicate that the approach is feasible, with tracking accuracies on the order of 5 m. Srikanth Saripalli, Gaurav S. Sukhatme	IROS	2
2002	Collective construction with multiple robots abstract We study the problem of construction by autonomous mobile robots focusing on the coordination strategy employed by the robots to solve a simple construction problem efficiently. In particular we address the problem of constructing a linear 2D structure in a planar bounded environment. A "minimalist" single-robot solution to the problem is given, as well as two multi-robot solutions, which are natural extensions to the single-robot approach, with varying degrees of inter-robot communication. Results show that with minimal inter-robot communication (1 bit of state), there is a significant improvement in the system performance. This improvement is invariant with respect to the size of the environment. Jens Wawerla, Gaurav S. Sukhatme, Maja J. Mataric	IROS	2
2002	LOST: localization-space trails for robot teams abstract We describe localization-space trails (LOST), a method that enables a team of robots to navigate between places of interest in an initially unknown environment using a trail of landmarks. The landmarks are not physical; they are waypoint coordinates generated online by each robot and shared with teammates. Waypoints are specified in each robot's local coordinate system, and contain references to features in the world that are relevant to the team's task and common to all robots. Using these task-level references, robots can share waypoints without maintaining a global coordinate system. The method is tested in a series of real-world multirobot experiments. The results demonstrate that the method: 1) copes with accumulating odometry error; 2) is robust to the failure of individual robots; 3) converges to the best route discovered by any robot in the team. In one experiment, a team of four autonomous mobile robots performs a resource transportation task in our uninstrumented office building. Despite significant divergence of their local coordinate systems, the robots are able to share waypoints, forming and following a common trail between two predetermined locations for more than three hours, traveling a total of 8.2 km (5.1 miles) before running out of power. Designed to scale to large populations, LOST is fully distributed, with low costs in processing, memory, and bandwidth. It combines metric data about the position of features in the world with instructions on how to get from one place to another; producing something between a map and a plan. Richard Vaughan 0001, Kasper Støy, Gaurav S. Sukhatme, Maja J. Mataric	IEEE Trans. Robotics Autom.	3
2001	Evaluating Control Strategies for Wireless-Networked Robots Using an Integrated Robot and Network Simulation abstract Wireless communication is an enabling factor in multiple mobile robot systems. There is significant interaction between robot controllers and communications subsystems. We present a method for evaluating combined robot control/communication strategies for a team of wireless-networked robots performing a resource transportation task. Two alternative controller designs are compared under established communication and radio propagation models. For each we measure the overall performance of the robot team including the cost of communication. The study illustrates how our evaluation tools can be used for designing controllers for robots operating in wireless communication environments. Wei Ye 0003, Richard Vaughan 0001, Gaurav S. Sukhatme, John S. Heidemann	ICRA	3
2001	Most valuable player: a robot device server for distributed control abstract Successful distributed sensing and control require data to flow effectively between sensors, processors and actuators on single robots, in groups and across the Internet. We propose a mechanism for achieving this flow that we have found to be powerful and easy to use; we call it Player. Player combines an efficient message protocol with a simple device model. It is implemented as a multithreaded TCP socket server that provides transparent network access to a collection of sensors and actuators, often comprising a robot. The socket abstraction enables platform- and language-independent control of these devices, allowing the system designer to use the best tool for the task at hand Player is freely available from http://robotics.usc.edu/player. Brian P. Gerkey, Richard Vaughan 0001, Kasper Støy, Andrew Howard 0001, Gaurav S. Sukhatme, Maja J. Mataric	IROS	5
2001	Relaxation on a mesh: a formalism for generalized localization abstract This paper considers two problems which at first sight appear to be quite distinct: localizing a robot in an unknown environment and calibrating an embedded sensor network. We show that both of these can be formulated as special cases of a generalized localization problem. In the standard localization problem, the aim is to determine the pose of some object (usually a mobile robot) relative to a global coordinate system. In our generalized version, the aim is to determine the pose of all elements in a network (both fixed and mobile) relative to an arbitrary global coordinate system. We have developed a physically inspired 'mesh-based' formalism for solving such problems. This paper outlines the formalism, and describes its application to the concrete tasks of multirobot mapping and calibration of a distributed sensor network. The paper presents experimental results for both tasks obtained using a set of Pioneer mobile robots equipped with scanning laser range-finders. Andrew Howard 0001, Maja J. Mataric, Gaurav S. Sukhatme	IROS	3
2001	Distributed multi-robot task allocation for emergency handling abstract We describe a prototype task, emergency handling, for multi-robot coordination. The experiments reported measure the effects of individualism and opportunism in a physically-implemented multi-robot system. We use sound at multiple frequencies to simulate emergencies by producing several locally-sensable gradients in the environment. Our results show that opportunism affords a significant performance improvement over individualism. Our experiments also demonstrate the viability of sound for producing detectable local gradients in the environment. Esben Hallundbæk Østergaard, Maja J. Mataric, Gaurav S. Sukhatme	IROS	3
2000	Fault Detection and Identification in a Mobile Robot using Multiple Model Estimation and Neural Network abstract We propose a method to detect and identify faults in wheeled mobile robots. The idea behind the method is to use adaptive estimation to predict the outcome of several faults, and to learn them collectively as a failure pattern. Models of the system behavior under each type of fault are embedded in multiple parallel Kalman filter (KF) estimators. Each KF is tuned to a particular fault and predicts, using its embedded model, the expected values for the sensor readings. The residual, the difference between the predicted readings (based on certain assumptions for the system model and the sensor models) and the actual sensor readings, is used as an indicator of how well each filter is performing. A backpropagation neural network processes this set of residuals as a pattern and decides which fault has occurred, that is, which filter is better tuned to the correct state of the mobile robot. The technique has been implemented on a physical robot and results from experiments are discussed. Puneet Goel, Göksel Dedeoglu, Stergios I. Roumeliotis, Gaurav S. Sukhatme	ICRA	4
2000	Sonar-based feature recognition and robot navigation using a neural network abstract An efficient strategy based on an artificial neural network is presented to enable a mobile robot to move autonomously and build a rough map of its environment. The paper describes a technique based on a neural network which uses data from sonars to recognize features online. Our method is systematic, simple and yields very good results. Moreover, it is generalizable in the sense that one can train the robot to recognize particular features of interest. We have implemented and tested this technique successfully on-board a mobile robot which uses a 68332 processor and does not have support for floating point arithmetic and file structure. Once trained, the robot can recognize the features on the fly and react to them appropriately. Puneet Goel, Gaurav S. Sukhatme	IROS	2
1999	Circumventing Dynamic Modeling: Evaluation of the Error-State Kalman Filter Applied to Mobile Robot Localization abstract The mobile robot localization problem is treated as a two-stage iterative estimation process. The attitude is estimated first and is then available for position estimation. The indirect (error state) form of the Kalman filter is developed for attitude estimation when applying gyro modeling. The main benefit of this choice is that combined dynamic modeling of the mobile robot and its interaction with the environment is avoided. The filter optimally combines the attitude rate information from the gyro and the absolute orientation measurements. The proposed implementation is independent of the structure of the vehicle or the morphology of the ground. The method can easily be transferred to another mobile platform provided it carries an equivalent set of sensors. The 2D case is studied in detail first. Results of extending the approach to the 3D case are presented. In both cases the results demonstrate the efficacy of the proposed method. Stergios I. Roumeliotis, Gaurav S. Sukhatme, George A. Bekey	ICRA	2
1999	Smoother Based 3-D Attitude Estimation for Mobile Robot Localization abstract The mobile robot localization problem is decomposed into two stages; attitude estimation followed by position estimation. The innovation of our method is the use of a smoother, in the attitude estimation loop that outperforms other Kalman filter based techniques in estimate accuracy. The smoother exploits the special nature of the data fused; high frequency inertial sensor (gyroscope) data and low frequency absolute orientation data (from a compass or sun sensor). Two Kalman filters form the smoother. During each time interval one of them propagates the attitude estimate forward in time until it is updated by an absolute orientation sensor. At this time, the second filter propagates the recently renewed estimate back in time. The smoother optimally exploits the limited observability of the system by combining the outcome of the two filters. The system model uses gyro modeling which relies on integrating the kinematic equations to propagate the attitude estimates and obviates the need for complex dynamic modeling. The indirect (error state) form of the Kalman filter is developed for both parts of the smoother. The proposed approach is independent of the robot structure and the morphology of the ground. It can easily be transferred to another robot which has an equivalent set of sensors. Quaternions are used for the 3D attitude representation mainly for practical reasons discussed in the paper. Proposed innovative algorithm is tested in simulation and the overall improvement in position estimation is demonstrated. Stergios I. Roumeliotis, Gaurav S. Sukhatme, George A. Bekey	ICRA	2
1999	Robust localization using relative and absolute position estimates abstract A low cost strategy based on well calibrated odometry is presented for localizing mobile robots. The paper describes a two-step process for correction of 'systematic errors' in encoder measurements followed by fusion of the calibrated odometry with a gyroscope and GPS resulting in a robust localization scheme. A Kalman filter operating on data from the sensors is used for estimating position and orientation of the robot. Experimental results are presented that show an improvement of at least one order of magnitude in accuracy compared to the un-calibrated, un-filtered case. Our method is systematic, simple and yields very good results. We show that this strategy proves useful when the robot is using GPS to localize itself as well as when GPS becomes unavailable for some time. As a result robot can move in and out of enclosed spaces, such as buildings, while keeping track of its position on the fly. Puneet Goel, Stergios I. Roumeliotis, Gaurav S. Sukhatme	IROS	3
1999	State estimation of an autonomous helicopter using Kalman filtering abstract Presents a technique to accurately estimate the state of a robot helicopter using a combination of gyroscopes, accelerometers, inclinometers and GPS. Simulation results of state estimation of the helicopter are presented using Kalman filtering based on sensor modeling. The number of estimated states of helicopter is nine : three attitudes(/spl theta/,/spl phi/,/spl psi/) from the gyroscopes, three accelerations(x/spl I.oarr/,y/spl I.oarr/,z/spl I.oarr/) and three positions (x, y, z) from the accelerometers. Two Kalman filters were used, one for the gyroscope data and the other for the accelerometer data. Our approach is unique because it explicitly avoids dynamic modeling of the system and allows for can elegant combination of sensor data available at different frequencies. We also describe the larger context in which this work is embedded, namely the design and implementation of an autonomous robot helicopter. Myungsoo Jun, Stergios I. Roumeliotis, Gaurav S. Sukhatme	IROS	3
1998	Fault Detection and Identification in a Mobile Robot Using Multiple-Model Estimation abstract This paper introduces a method to detect and identify faults in wheeled mobile robots. The idea behind the method is to use adaptive estimation to predict (in parallel) the outcome of several faults. Models of the system behavior under each type of fault are embedded in the various parallel estimators (each of which is a Kalman filter). Each filter is thus tuned to a particular fault. Using its embedded model each filter predicts values for the sensor readings. The residual (the difference between the predicted and actual sensor reading) is an indicator of how well the filter is performing. A fault detection and identification module is responsible for processing the residual to decide which fault has occurred. As an example the method is implemented successfully on a Pioneer I robot. The paper concludes with a discussion of future work. Stergios I. Roumeliotis, Gaurav S. Sukhatme, George A. Bekey	ICRA	2
1998	Sensor fault detection and identification in a mobile robot abstract Multiple model adaptive estimation (MMAE) is used to detect and identify sensor failures in a mobile robot. Each estimator is a Kalman filter with a specific embedded failure model. The filter bank also contains one filter which has the nominal model embedded within it. The filter residuals are postprocessed to produce a probabilistic interpretation of the operation of the system. The output of the system at any given time is the confidence in the correctness of the various embedded models. As an additional feature the standard assumption that the measurements are available at a constant, common frequency, is relaxed. Measurements are assumed to be asynchronous and of varying frequency. The particularly difficult case of 'soft' sensor failure is also handled successfully. A system architecture is presented for the general problem of failure detection and identification in mobile robots. As an example, the MMAE algorithm is demonstrated on a Pioneer I robot in the case of three different sensor failures. Stergios I. Roumeliotis, Gaurav S. Sukhatme, George A. Bekey	IROS	2
1997	Mobility evaluation of a wheeled microrover using a dynamic model abstract This paper describes a multiple criteria, statistical technique for mobile robot evaluation. The evaluation method measures the time and energy costs of a particular class of exploratory missions. The method is implemented in the special case of one wheeled mobile robot in the laboratory, which was made to execute several (approximately 100) instances of the mission. It is proposed that a validated, dynamic model of the robot embedded in a simulated mission scenario is to be used to study alternate robot designs in a small 'neighborhood' of the existing physical system. Such a simulation was built and is described here. Results from trials with the physical robot and its simulated counterpart are compared and shown to agree well. The simulation is used to evaluate designs not constructed in the laboratory, and results are discussed. Gaurav S. Sukhatme, Scott Brizius, George A. Bekey	IROS	1
1995	Mission Reachability for Extraterrestrial Rovers abstract A methodology for the mobility evaluation and mission oriented performance assessment of an autonomous mobile robot is developed. The evaluation strategy uses both time and energy based measures resulting in the development of isochronal/isoenergy contours which characterize the expected mobility of the robot over novel terrain using sample trials on patches of land prior to the mission. This work has a basis in the evaluation of a robot for extraterrestrial exploration and demonstrates the feasibility of the evaluation methodology proposed using simulations. Simulation results are reported for two designs of micro-rovers viz. legged and wheeled on a simulated Martian terrain. Future work is discussed in the context of increased complexity due to hardware implementations, better terrain models and statistical simulations all of which are currently under development. Gaurav S. Sukhatme, M. Anthony Lewis, George A. Bekey	ICRA	1
1994	On the Development of EMG Control for a Prosthesis Using a Robotic Hand abstract The human hand is a complex end-effector capable of a large variety of postures. Multifingered robot hands, such as the Belgrade/USC hand, can approximate human hand functionality, and it is possible to consider their use in prosthetics. The authors have developed a system, PRESHAPE, that translates user commands into motor signals using the virtual finger concept. For control, electromyographic (EMG) signals from forearm muscles are used. The authors describe PRESHAPE and its use of EMG signals. Simulation results are presented.> Thea Iberall, Gaurav S. Sukhatme, Denise Beattie, George A. Bekey	ICRA	2
1993	Control philosophy and simulation of a robotic hand as a model for prosthetic hands abstract Multi-fingered robotic hands are attempts to approximate human hand characteristics and functionality, and it is reasonable to consider their possible adaptation and use in prosthetics and rehabilitation. The Belgrade/USC robot hand is used as a prototype prosthetic hand in order to evaluate a system that translates task-level commands into motor commands. The system, PRESHAPE, uses the virtual finger concept for generating the free and guarded motions that occur during the phases of hand movements in prehensile and nonprehensile tasks. This paper describes the control philosophy of PRESHAPE and presents simulation results for various tasks. Thea Iberall, Gaurav S. Sukhatme, Denise Beattie, George A. Bekey	IROS	2