As humans and mobile robots increasingly coexist in public spaces, their close proximity demands that robots navigate following navigation strategies similar to those exhibited by humans. This could be achieved by learning directly from human demonstration trajectories in a machine learning framework. In this paper, we present a method to learn human navigation behaviors using an imitation learning approach based on generative adversarial imitation learning (GAIL), which has the ability of directly extracting navigation policy. Specifically, we use a large open human trajectory dataset that was experimentally collected in a crowded public space. We then recreate these human trajectories in a 3D robotic simulator, and generate demonstration data using a LIDAR sensor onboard a robot with the robot following the measured human trajectories. We then propose a GAIL based algorithm, which uses occupancy maps generated using LIDAR data as the input, and outputs the navigation policy for robot navigation. Simulation experiments are conducted, and performance evaluation shows that the learned navigation policy generates trajectories qualitatively and quantitatively similar to human trajectories. Compared with existing works using analytical models (such as social force model) to generate human demonstration trajectories, our method learns directly from intrinsic human trajectories, thus exhibits more human-like navigation behaviors.

Pedestrian regulation can prevent crowd accidents and improve crowd safety in densely populated areas. Recent studies use mobile robots to regulate pedestrian flows for desired collective motion through the effect of passive human-robot interaction (HRI). This paper formulates a robot motion planning problem for the optimization of two merging pedestrian flows moving through a bottleneck exit. To address the challenge of feature representation of complex human motion dynamics under the effect of HRI, we propose using a deep neural network to model the mapping from the image input of pedestrian environments to the output of robot motion decisions. The robot motion planner is trained end-to-end using a deep reinforcement learning algorithm, which avoids hand-crafted feature detection and extraction, thus improving the learning capability for complex dynamic problems. Our proposed approach is validated in simulated experiments, and its performance is evaluated. The results demonstrate that the robot is able to find optimal motion decisions that maximize the pedestrian outflow in different flow conditions, and the pedestrian-accumulated outflow increases significantly compared to cases without robot regulation and with random robot motion.

Humans and mobile robots will be increasingly cohabiting in the same environments, which has lead to an increase in studies on human robot interaction (HRI). One important topic in these studies is the development of robot navigation algorithms that are socially compliant to humans navigating in the same space. In this paper, we present a method to learn human navigation behaviors using maximum entropy deep inverse reinforcement learning (MEDIRL). We use a large open dataset of pedestrian trajectories collected in an uncontrolled environment as the expert demonstrations. Human navigation behaviors are captured by a nonlinear reward function through deep neural network (DNN) approximation. The developed MEDIRL algorithm takes feature inputs including social affinity map (SAM) that are extracted from human motion trajectories. We perform simulation experiments using the learned reward function, and the performance is evaluated comparing it with the real measured pedestrian trajectories in the dataset. The evaluation results show that the proposed method has acceptable prediction accuracy compared to other state-of-the-art methods, and it can generate pedestrian trajectories similar to real human trajectories with natural social navigation behaviors such as collision avoidance, leader-follower, and split-and-rejoin.

We present field experiment results of ocean plume surveys conducted with a robotic platform. In our experiments, Rhodamine dye was used to generate a chemical plume in a coastal environment, and an unmanned surface vessel equipped with fluorometer sensors was deployed to conduct plume surveys. We present plume front and cross sectional survey results, where the robot is driven along visible plume boundaries and plume cross sections, respectively. Datasets collected during the experiments are shown together with statistical parameters that quantify the fine-scale plume structure. Experimental evidence shows that the assumption of the existence of a continuous and fixed-value concentration curve used in existing theoretical work for plume front tracking appears too strong for real ocean plumes. The datasets provide characteristic parameters that can be used for further development and verification of simulation models of ocean pollution plumes.

Using autonomous mobile robots to dynamically track oil plume propagation in ocean environments is challenging. Based on a model of advection-diffusion equation that describes point-source pollution propagation in marine environments, we have previously proposed a model-based estimator-controller design for autonomous robots to dynamically track plume front propagation. In this paper we study the robustness of the controller in a robot simulator, and evaluate it's performance in a realistic environmental model setup. A probabilistic Lagrangian environmental model is used that can capture both the time-averaged, idealized structure and the instantaneous, realistic structure of a dynamic plume. The controller is implemented on the Field Robotics Lab Vehicle Software, which uses Lightweight Communication and Marshalling library to facilitate process communication and for easy transit to field testing. Simulation results are shown with discussions and lessons learned to guide future field experiments.

Human indoor localization was previously implemented using wireless sensor networks at the cost of sensing infrastructure deployment. Motivated by high density of smartphones in public spaces, we propose to use a robot-assisted localization system in which the low-cost Kinect sensor and smartphone-based acoustic relative ranging are used to localize moving human targets in indoor environments. An extended Kalman filter based localization algorithm is developed for real-time dynamic position estimation. We present both simulations and real robot-smartphone experiments demonstrating the performance with a localization accuracy of approximately 0.5m.