We introduce Scenario Dreamer, a fully data-driven generative simulator for autonomous vehicle planning that generates both the initial traffic scene—comprising a lane graph and agent bounding boxes—and closed-loop agent behaviours. Existing methods for generating driving simulation environments encode the initial traffic scene as a rasterized image and, as such, require parameter-heavy networks that perform unnecessary computation due to many empty pixels in the rasterized scene. Moreover, we find that existing methods that employ rule-based agent behaviours lack diversity and realism. Scenario Dreamer instead employs a novel vectorized latent diffusion model for initial scene generation that directly operates on the vectorized scene elements and an autoregressive Transformer for data-driven agent behaviour simulation. Scenario Dreamer additionally supports scene extrapolation via diffusion inpainting, enabling the generation of unbounded simulation environments. Extensive experiments show that Scenario Dreamer outperforms existing generative simulators in realism and efficiency: the vectorized scene-generation base model achieves superior generation quality with around 2× fewer parameters, 6× lower generation latency, and 10× fewer GPU training hours compared to the strongest baseline. We confirm its practical utility by showing that reinforcement learning planning agents are more challenged in Scenario Dreamer environments than traditional non-generative simulation environments, especially on long and adversarial driving environments.

Reinforcement learning algorithms typically necessitate extensive exploration of the state space to find optimal policies. However, in safety-critical applications, the risks associated with such exploration can lead to catastrophic consequences. Existing safe exploration methods attempt to mitigate this by imposing constraints, which often result in overly conservative behaviours and inefficient learning. Heavy penalties for early constraint violations can trap agents in local optima, deterring exploration of risky yet high-reward regions of the state space. To address this, we introduce a method that explicitly learns state-conditioned safety representations. By augmenting the state features with these safety representations, our approach naturally encourages safer exploration without being excessively cautious, resulting in more efficient and safer policy learning in safety-critical scenarios. Empirical evaluations across diverse environments show that our method significantly improves task performance while reducing constraint violations during training, underscoring its effectiveness in balancing exploration with safety.

Many robotic systems require extended deployments in complex, dynamic environments. In such deployments, parts of the environment may change between subsequent robot observations. Most robotic mapping or environment modeling algorithms are incapable of representing dynamic features in a way that enables predicting their future state. Instead, they opt to filter certain state observations, either by removing them or some form of weighted averaging. This paper introduces Perpetua, a method for modeling the dynamics of semi-static features. Perpetua is able to: incorporate prior knowledge about the dynamics of the feature if it exists, track multiple hypotheses, and adapt over time to enable predicting of future feature states. Specifically, we chain together mixtures of "persistence" and "emergence" filters to model the probability that features will disappear or reappear in a formal Bayesian framework. The approach is an efficient, scalable, general, and robust method for estimating the states of features in an environment, both in the present as well as at arbitrary future times. Through experiments on simulated and real-world data, we find that Perpetua yields better accuracy than similar approaches while also being online adaptable and robust to missing observations.

3D scene understanding has been transformed by open-vocabulary language models that enable interaction via natural language. However, at present the evaluation of these representations is limited to datasets with closed-set semantics that do not capture the richness of language. This work presents OpenLex3D, a dedicated benchmark for evaluating 3D open-vocabulary scene representations. OpenLex3D provides entirely new label annotations for scenes from Replica, ScanNet++, and HM3D, which capture real-world linguistic variability by introducing synonymical object categories and additional nuanced descriptions. Our label sets provide 13 times more labels per scene than the original datasets. By introducing an open-set 3D semantic segmentation task and an object retrieval task, we evaluate various existing 3D open-vocabulary methods on OpenLex3D, showcasing failure cases, and avenues for improvement. Our experiments provide insights on feature precision, segmentation, and downstream capabilities. The benchmark is publicly available at: https://openlex3d.github.io/.

The creation of photorealistic virtual worlds requires the accurate modeling of 3D surface geometry for a wide range of objects. For this, meshes are appealing since they enable 1) fast physics-based rendering with realistic material and lighting, 2) physical simulation, and 3) are memory-efficient for modern graphics pipelines. Recent work on reconstructing and statistically modeling 3D shape, however, has critiqued meshes as being topologically inflexible. To capture a wide range of object shapes, any 3D representation must be able to model solid, watertight, shapes as well as thin, open, surfaces. Recent work has focused on the former, and methods for reconstructing open surfaces do not support fast reconstruction with material and lighting or unconditional generative modelling. Inspired by the observation that open surfaces can be seen as islands floating on watertight surfaces, we parametrize open surfaces by defining a manifold signed distance field on watertight templates. With this parametrization, we further develop a grid-based and differentiable representation that parametrizes both watertight and non-watertight meshes of arbitrary topology. Our new representation, called Ghost-on-the-Shell (G-Shell), enables two important applications: differentiable rasterization-based reconstruction from multiview images and generative modelling of non-watertight meshes. We empirically demonstrate that G-Shell achieves state-of-the-art performance on non-watertight mesh reconstruction and generation tasks, while also performing effectively for watertight meshes.

For robots to perform a wide variety of tasks, they require a 3D representation of the world that is semantically rich, yet compact and efficient for task-driven perception and planning. Recent approaches have attempted to leverage features from large vision-language models to encode semantics in 3D representations. However, these approaches tend to produce maps with per-point feature vectors, which do not scale well in larger environments, nor do they contain semantic spatial relationships between entities in the environment, which are useful for downstream planning. In this work, we propose ConceptGraphs, an open-vocabulary graph-structured representation for 3D scenes. ConceptGraphs is built by leveraging 2D foundation models and fusing their output to 3D by multi-view association. The resulting representations generalize to novel semantic classes, without the need to collect large 3D datasets or finetune models. We demonstrate the utility of this representation through a number of downstream planning tasks that are specified through abstract (language) prompts and require complex reasoning over spatial and semantic concepts. To explore the full scope of our experiments and results, we encourage readers to visit our project webpage.

Estimating the uncertainty in deep neural network predictions is crucial for many real-world applications. A common approach to model uncertainty is to choose a parametric distribution and fit the data to it using maximum likelihood estimation. The chosen parametric form can be a poor fit to the data-generating distribution, resulting in unreliable uncertainty estimates. In this work, we propose SampleNet, a flexible and scalable architecture for modeling uncertainty that avoids specifying a parametric form on the output distribution. SampleNets do so by defining an empirical distribution using samples that are learned with the Energy Score and regularized with the Sinkhorn Divergence. SampleNets are shown to be able to well-fit a wide range of distributions and to outperform baselines on large-scale real-world regression tasks.

An effective framework for learning 3D representations for perception tasks is distilling rich self-supervised image features via contrastive learning. However, image-to-point representation learning for autonomous driving datasets faces two main challenges: 1) the abundance of self-similarity, which results in the contrastive losses pushing away semantically similar point and image regions and thus disturbing the local semantic structure of the learned representations, and 2) severe class imbalance as pretraining gets dominated by over-represented classes. We propose to alleviate the self-similarity problem through a novel semantically tolerant image-to-point contrastive loss that takes into consideration the semantic distance between positive and negative image regions to minimize contrasting semantically similar point and image regions. Additionally, we address class imbalance by designing a class-agnostic balanced loss that approximates the degree of class imbalance through an aggregate sample-to-samples semantic similarity measure. We demonstrate that our semantically-tolerant contrastive loss with class balancing improves state-of-the-art 2D-to-3D representation learning in all evaluation settings on 3D semantic segmentation. Our method consistently outperforms state-of-the-art 2D-to-3D representation learning frameworks across a wide range of 2D self-supervised pretrained models.

A fundamental task in robotics is to navigate between two locations. In particular, real-world navigation can require long-horizon planning using high-dimensional RGB images, which poses a substantial challenge for end-to-end learning-based approaches. Current semi-parametric methods instead achieve long-horizon navigation by combining learned modules with a topological memory of the environment, often represented as a graph over previously collected images. However, using these graphs in practice requires tuning a number of pruning heuristics. These heuristics are necessary to avoid spurious edges, limit runtime memory usage and maintain reasonably fast graph queries in large environments. In this work, we present One-4-All (O4A), a method leveraging self-supervised and manifold learning to obtain a graph-free, end-to-end navigation pipeline in which the goal is specified as an image. Navigation is achieved by greedily minimizing a potential function defined continuously over image embeddings. Our system is trained offline on non-expert exploration sequences of RGB data and controls, and does not require any depth or pose measurements. We show that 04A can reach long-range goals in 8 simulated Gibson indoor environments and that resulting embeddings are topologically similar to ground truth maps, even if no pose is observed. We further demonstrate successful real-world navigation using a Jackal UGV platform.aaProject page https://montrealrobotics.ca/o4a/.

While modern deep neural networks are performant perception modules, performance (accuracy) alone is insufficient, particularly for safety-critical robotic applications such as self-driving vehicles. Robot autonomy stacks also require these otherwise blackbox models to produce reliable and calibrated measures of confidence on their predictions. Existing approaches estimate uncertainty from these neural network perception stacks by modifying network architectures, inference procedure, or loss functions. However, in general, these methods lack calibration, meaning that the predictive uncertainties do not faithfully represent the true underlying uncertainties (process noise). Our key insight is that calibration is only achieved by imposing constraints across multiple examples, such as those in a mini-batch; as opposed to existing approaches which only impose constraints per-sample, often leading to overconfident (thus miscalibrated) uncertainty estimates. By enforcing the distribution of outputs of a neural network to resemble a target distribution by minimizing an$f$-divergence, we obtain significantly better-calibrated models compared to prior approaches. Our approach, f-Cal, outperforms existing uncertainty calibration approaches on robot perception tasks such as object detection and monocular depth estimation over multiple real-world benchmarks.

The inductive bias of a neural network is largely determined by the architecture and the training algorithm. To achieve good generalization, how to effectively train a neural network is of great importance. We propose a novel orthogonal over-parameterized training (OPT) framework that can provably minimize the hyperspherical energy which characterizes the diversity of neurons on a hypersphere. By maintaining the minimum hyperspherical energy during training, OPT can greatly improve the empirical generalization. Specifically, OPT fixes the randomly initialized weights of the neurons and learns an orthogonal transformation that applies to these neurons. We consider multiple ways to learn such an orthogonal transformation, including unrolling orthogonalization algorithms, applying orthogonal parameterization, and designing orthogonality-preserving gradient descent. For better scalability, we propose the stochastic OPT which performs orthogonal transformation stochastically for partial dimensions of neurons. Interestingly, OPT reveals that learning a proper coordinate system for neurons is crucial to generalization. We provide some insights on why OPT yields better generalization. Extensive experiments validate the superiority of OPT over the standard training.

In many situations it is either impossible or impractical to develop and evaluate agents entirely on the target domain on which they will be deployed. This is particularly true in robotics, where doing experiments on hardware is much more arduous than in simulation. This has become arguably more so in the case of learning-based agents. To this end, considerable recent effort has been devoted to developing increasingly realistic and higher fidelity simulators. However, we lack any principled way to evaluate how good a "proxy domain" is, specifically in terms of how useful it is in helping us achieve our end objective of building an agent that performs well in the target domain. In this work, we investigate methods to address this need. We begin by clearly separating two uses of proxy domains that are often conflated: 1) their ability to be a faithful predictor of agent performance and 2) their ability to be a useful tool for learning. In this paper, we attempt to clarify the role of proxy domains and establish new proxy usefulness (PU) metrics to compare the usefulness of different proxy domains. We propose the relative predictive PU to assess the predictive ability of a proxy domain and the learning PU to quantify the usefulness of a proxy as a tool to generate learning data. Furthermore, we argue that the value of a proxy is conditioned on the task that it is being used to help solve. We demonstrate how these new metrics can be used to optimize parameters of the proxy domain for which obtaining ground truth via system identification is not trivial.

In this paper, we consider the problem of iterative machine teaching, where a teacher provides examples sequentially based on the current iterative learner. In contrast to previous methods that have to scan over the entire pool and select teaching examples from it in each iteration, we propose a label synthesis teaching framework where the teacher randomly selects input teaching examples (e.g., images) and then synthesizes suitable outputs (e.g., labels) for them. We show that this framework can avoid costly example selection while still provably achieving exponential teachability. We propose multiple novel teaching algorithms in this framework. Finally, we empirically demonstrate the value of our framework.

Modern generative models are usually designed to match target distributions directly in the data space, where the intrinsic dimension of data can be much lower than the ambient dimension. We argue that this discrepancy may contribute to the difficulties in training generative models. We therefore propose to map both the generated and target distributions to the latent space using the encoder of a standard autoencoder, and train the generator (or decoder) to match the target distribution in the latent space. Specifically, we enforce the consistency in both the data space and the latent space with theoretically justified data and latent reconstruction losses. The resulting generative model, which we call a perceptual generative autoencoder (PGA), is then trained with a maximum likelihood or variational autoencoder (VAE) objective. With maximum likelihood, PGAs generalize the idea of reversible generative models to unrestricted neural network architectures and arbitrary number of latent dimensions. When combined with VAEs, PGAs substantially improve over the baseline VAEs in terms of sample quality. Compared to other autoencoder-based generative models using simple priors, PGAs achieve state-of-the-art FID scores on CIFAR-10 and CelebA.

The question of "representation" is central in the context of dense simultaneous localization and mapping (SLAM). Learning-based approaches have the potential to leverage data or task performance to directly inform the representation. However, blending representation learning approaches with "classical" SLAM systems has remained an open question, because of their highly modular and complex nature. A SLAM system transforms raw sensor inputs into a distribution over the state(s) of the robot and the environment. If this transformation (SLAM) were expressible as a differentiable function, we could leverage task-based error signals over the outputs of this function to learn representations that optimize task performance. However, this is infeasible as several components of a typical dense SLAM system are non-differentiable. In this work, we propose ∇SLAM (gradSLAM), a methodology for posing SLAM systems as differentiable computational graphs, which unifies gradient-based learning and SLAM. We propose differentiable trust-region optimizers, surface measurement and fusion schemes, and raycasting, without sacrificing accuracy. This amalgamation of dense SLAM with computational graphs enables us to backprop all the way from 3D maps to 2D pixels, opening up new possibilities in gradient-based learning for SLAM1.

As robotics matures and increases in complexity, it is more necessary than ever that robot autonomy research be reproducible. Compared to other sciences, there are specific challenges to benchmarking autonomy, such as the complexity of the software stacks, the variability of the hardware and the reliance on data-driven techniques, amongst others. In this paper, we describe a new concept for reproducible robotics research that integrates development and benchmarking, so that reproducibility is obtained "by design" from the beginning of the research/development processes. We first provide the overall conceptual objectives to achieve this goal and then a concrete instance that we have built: the DUCKIENet. One of the central components of this setup is the Duckietown Autolab, a remotely accessible standardized setup that is itself also relatively low-cost and reproducible. When evaluating agents, careful definition of interfaces allows users to choose among local versus remote evaluation using simulation, logs, or remote automated hardware setups. We validate the system by analyzing the repeatability of experiments conducted using the infrastructure and show that there is low variance across different robot hardware and across different remote labs.†

We show that the sum of the implicit generator log-density $\log p_g$ of a GAN with the logit score of the discriminator defines an energy function which yields the true data density when the generator is imperfect but the discriminator is optimal, thus making it possible to improve on the typical generator (with implicit density $p_g$). To make that practical, we show that sampling from this modified density can be achieved by sampling in latent space according to an energy-based model induced by the sum of the latent prior log-density and the discriminator output score. This can be achieved by running a Langevin MCMC in latent space and then applying the generator function, which we call Discriminator Driven Latent Sampling~(DDLS). We show that DDLS is highly efficient compared to previous methods which work in the high-dimensional pixel space and can be applied to improve on previously trained GANs of many types. We evaluate DDLS on both synthetic and real-world datasets qualitatively and quantitatively. On CIFAR-10, DDLS substantially improves the Inception Score of an off-the-shelf pre-trained SN-GAN~\citep{sngan} from $8.22$ to $9.09$ which is even comparable to the class-conditional BigGAN~\citep{biggan} model. This achieves a new state-of-the-art in unconditional image synthesis setting without introducing extra parameters or additional training.

The continual learning problem involves training models with limited capacity to perform well on a set of an unknown number of sequentially arriving tasks. While meta-learning shows great potential for reducing interference between old and new tasks, the current training procedures tend to be either slow or offline, and sensitive to many hyper-parameters. In this work, we propose Look-ahead MAML (La-MAML), a fast optimisation-based meta-learning algorithm for online-continual learning, aided by a small episodic memory. By incorporating the modulation of per-parameter learning rates in our meta-learning update, our approach also allows us to draw connections to and exploit prior work on hypergradients and meta-descent. This provides a more flexible and efficient way to mitigate catastrophic forgetting compared to conventional prior-based methods. La-MAML achieves performance superior to other replay-based, prior-based and meta-learning based approaches for continual learning on real-world visual classification benchmarks.

Learning effective visuomotor policies for robots purely from data is challenging, but also appealing since a learning-based system should not require manual tuning or calibration. In the case of a robot operating in a real environment the training process can be costly, time-consuming, and even dangerous since failures are common at the start of training. For this reason, it is desirable to be able to leverage simulation and off-policy data to the extent possible to train the robot. In this work, we introduce a robust framework that plans in simulation and transfers well to the real environment. Our model incorporates a gradient-descent based planning module, which, given the initial image and goal image, encodes the images to a lower dimensional latent state and plans a trajectory to reach the goal. The model, consisting of the encoder and planner modules, is first trained through a meta-learning strategy in simulation. We subsequently perform adversarial domain transfer on the encoder by using a bank of unlabelled but random images from the simulation and real environments to enable the encoder to map images from the real and simulated environments to a similarly distributed latent representation. By fine tuning the entire model (encoder + planner) with only a few real world expert demonstrations, we show successful planning performances in different navigation tasks.

Deep learning has been successfully applied to “end-to-end” learning of the autonomous driving task, where a deep neural network learns to predict steering control commands from camera data input. However, the learned representations do not support higher-level decision making required for autonomous navigation, nor the uncertainty estimates required for parallel autonomy, where vehicle control is shared between human and robot. This paper tackles the problem of learning a representation to predict a continuous control probability distribution, and thus steering control options and bounds for those options, which can be used for autonomous navigation. Each mode of the distribution encodes a possible macro-action that the system could execute at that instant, and the covariances of the modes place bounds on safe steering control values. Our approach has the added advantage of being trained on unlabeled data collected from inexpensive cameras. The deep neural network based algorithm generates a probability distribution over the space of steering angles, from which we leverage Variational Bayesian methods to extract a mixture model and compute the different possible actions in the environment. A bound, which the autonomous vehicle must respect in our parallel autonomy setting, is then computed for each of these actions. We evaluate our approach on a challenging dataset containing a wide variety of driving conditions, and show that our algorithm is capable of parameterizing Gaussian Mixture Models for possible actions, and extract steering bounds with a mean error of only 2 degrees. Additionally, we demonstrate our system working on a full scale autonomous vehicle and evaluate its ability to successful handle various different parallel autonomy situations.

State-of-the-art autonomous driving systems rely heavily on detailed and highly accurate prior maps. However, outside of small urban areas, it is very challenging to build, store, and transmit detailed maps since the spatial scales are so large. Furthermore, maintaining detailed maps of large rural areas can be impracticable due to the rapid rate at which these environments can change. This is a significant limitation for the widespread applicability of autonomous driving technology, which has the potential for an incredibly positive societal impact. In this paper, we address the problem of autonomous navigation in rural environments through a novel mapless driving framework that combines sparse topological maps for global navigation with a sensor-based perception system for local navigation. First, a local navigation goal within the sensor view of the vehicle is chosen as a waypoint leading towards the global goal. Next, the local perception system generates a feasible trajectory in the vehicle frame to reach the waypoint while abiding by the rules of the road for the segment being traversed. These trajectories are updated to remain in the local frame using the vehicle's odometry and the associated uncertainty based on the least-squares residual and a recursive filtering approach, which allows the vehicle to navigate road networks reliably, and at high speed, without detailed prior maps. We demonstrate the performance of the system on a full-scale autonomous vehicle navigating in a challenging rural environment and benchmark the system on a large amount of collected data.

High-end vehicles are already equipped with safety systems, such as assistive braking and automatic lane following, enhancing vehicle safety. Yet, these current solutions can only help in low-complexity driving situations. In this paper, we introduce a parallel autonomy, or shared control, framework that computes safe trajectories for an automated vehicle, based on human inputs. We minimize the deviation from the human inputs while ensuring safety via a set of collision avoidance constraints. Our method achieves safe motion even in complex driving scenarios, such as those commonly encountered in an urban setting. We introduce a receding horizon planner formulated as nonlinear model predictive control (NMPC), which includes the analytic descriptions of road boundaries and the configuration and future uncertainties of other road participants. The NMPC operates over both steering and acceleration simultaneously. We introduce a nonslip model suitable for handling complex environments with dynamic obstacles, and a nonlinear combined slip vehicle model including normal load transfer capable of handling static environments. We validate the proposed approach in two complex driving scenarios. First, in an urban environment that includes a left-turn across traffic and passing on a busy street. And second, under snow conditions on a race track with sharp turns and under complex dynamic constraints. We evaluate the performance of the method with various human driving styles. We consequently observe that the method successfully avoids collisions and generates motions with minimal intervention for parallel autonomy. We note that the method can also be applied to generate safe motion for fully autonomous vehicles.

Duckietown is an open, inexpensive and flexible platform for autonomy education and research. The platform comprises small autonomous vehicles (“Duckiebots”) built from off-the-shelf components, and cities (“Duckietowns”) complete with roads, signage, traffic lights, obstacles, and citizens (duckies) in need of transportation. The Duckietown platform offers a wide range of functionalities at a low cost. Duckiebots sense the world with only one monocular camera and perform all processing onboard with a Raspberry Pi 2, yet are able to: follow lanes while avoiding obstacles, pedestrians (duckies) and other Duckiebots, localize within a global map, navigate a city, and coordinate with other Duckiebots to avoid collisions. Duckietown is a useful tool since educators and researchers can save money and time by not having to develop all of the necessary supporting infrastructure and capabilities. All materials are available as open source, and the hope is that others in the community will adopt the platform for education and research.

Current state-of-the-art vehicle safety systems, such as assistive braking or automatic lane following, are still only able to help in relatively simple driving situations. We introduce a Parallel Autonomy shared-control framework that produces safe trajectories based on human inputs even in much more complex driving scenarios, such as those commonly encountered in an urban setting. We minimize the deviation from the human inputs while ensuring safety via a set of collision avoidance constraints. We develop a receding horizon planner formulated as a Non-linear Model Predictive Control (NMPC) including analytic descriptions of road boundaries, and the configurations and future uncertainties of other traffic participants, and directly supplying them to the optimizer without linearization. The NMPC operates over both steering and acceleration simultaneously. Furthermore, the proposed receding horizon planner also applies to fully autonomous vehicles. We validate the proposed approach through simulations in a wide variety of complex driving scenarios such as left-turns across traffic, passing on busy streets, and under dynamic constraints in sharp turns on a race track.

Modern autonomous systems such as driverless vehicles need to safely operate in a wide range of conditions. A potential solution is to employ a hybrid systems approach, where safety is guaranteed in each individual mode within the system. This offsets complexity and responsibility from the individual controllers onto the complexity of determining discrete mode transitions. In this work we propose an efficient framework based on recursive neural networks and coreset data summarization to learn the transitions between an arbitrary number of controller modes that can have arbitrary complexity. Our approach allows us to efficiently gather annotation data from the large-scale datasets that are required to train such hybrid nonlinear systems to be safe under all operating conditions, favoring underexplored parts of the data. We demonstrate the construction of the embedding, and efficient detection of switching points for autonomous and non-autonomous car data. We further show how our approach enables efficient sampling of training data, to further improve either our embedding or the controllers.

We present the development of a full-scale “parallel autonomy” research platform including software and hardware. In the parallel autonomy paradigm, the control of the vehicle is shared; the human is still in control of the vehicle, but the autonomy system is always running in the background to prevent accidents. Our holistic approach includes: (1) a drive-by-wire conversion method only based on reverse engineering mounting of relatively inexpensive sensors onto the vehicle implementation of a localization and mapping system, (4) obstacle detection and (5) a shared controller as well as (6) integration with an advanced autonomy simulation system (Drake) for rapid development and testing. The system can operate in three modes: (a) manual driving, (b) full autonomy, where the system is in complete control of the vehicle and (c) parallel autonomy, where the shared controller is implemented. We present results from extensive testing of a full-scale vehicle on closed tracks that demonstrate these capabilities.

The operation of mobile robots in unknown environments typically requires building maps during exploration. As the exploration time and environment size increase, the amount of data collected and the number of variables required to represent these maps both grow, which is problematic since all real robots have finite resources. The solution proposed in this paper is to only retain the variables and measurements that are most important to achieve the robot's task. The variable and measurement selection approach is demonstrated on the task of navigation with a low risk of collision. Our approach has two stages: first, a subset of the variables is selected that is most useful for minimizing the uncertainty of navigation (termed the “focused variables”). And second, a task-agnostic method is used to select a subset of the measurements that maximizes the information over these focused variables (“focused inference”). Detailed simulations and hardware experiments show that the two-stage approach constrains the number of variables and measurements. It can generate much sparser maps than existing approaches in the literature, while still achieving a better task performance-in this case (fewer collisions). An incremental and iterative approach is further presented, in which the two-stage procedure is performed on subsets of the data, and thus, avoids the necessity of performing a resource-intensive batch selection on large datasets.

Graphical methods have proven an extremely useful tool employed by the mobile robotics community to frame estimation problems. Incremental solvers are able to process incoming sensor data and produce maximum a posteriori (MAP) estimates in realtime by exploiting the natural sparsity within the graph for reasonable-sized problems. However, to enable truly longterm operation in prior unknown environments requires algorithms whose computation, memory, and bandwidth (in the case of distributed systems) requirements scale constantly with time and environment size. Some recent approaches have addressed this problem through a two-step process - first the variables selected for removal are marginalized which induces density, and then the result is sparsified to maintain computational efficiency. Previous literature generally addresses only one of these two components. In this work, we attempt to explicitly connect all of the aforementioned resource constraint requirements by considering the node removal and sparsification pipeline in its entirety. We formulate the node selection problem as a minimization problem over the penalty to be paid in the resulting sparsification. As a result, we produce node subset selection strategies that are optimal in terms of minimizing the impact, in terms of Kullback-Liebler divergence (KLD), of approximating the dense distribution by a sparse one. We then show that one instantiation of this problem yields a computationally tractable formulation. Finally, we evaluate the method on standard datasets and show that the KLD is minimized as compared to other commonly-used heuristic node selection techniques.

Graph-based SLAM approaches have had success recently despite suffering from ever-increasing computational costs due to the need of optimizing over the entire robot trajectory. To address this issue, in this paper, we advocate the decoupling of marginalization (node removal) and sparsification (edge reduction) to allow for short-term retention of dense factors induced by marginalization while enabling us to spread the computation of these two operations over time. In particular, we analytically show that during marginalization, the correct choice of linearization points in constructing dense marginal factors is to use the relative (local), instead of global, state estimates in the Markov blanket of the marginalized node, which has lacked a general consensus in the literature. Furthermore, during sparsification, we determine an online sparse topology through sparsity-regularized convex optimization, which guides us to construct consistent sparse factors to best approximate the original dense factors across the Markov blanket. The proposed approach is tested extensively on both 2D and 3D public datasets and shown to perform competitively to the state-of-the-art algorithms.

Mapping and self-localization in unknown environments are fundamental capabilities in many robotic applications. These tasks typically involve the identification of objects as unique features or landmarks, which requires the objects both to be detected and then assigned a unique identifier that can be maintained when viewed from different perspectives and in different images. The data association and simultaneous localization and mapping (SLAM) problems are, individually, well-studied in the literature. But these two problems are inherently tightly coupled, and that has not been well-addressed. Without accurate SLAM, possible data associations are combinatorial and become intractable easily. Without accurate data association, the error of SLAM algorithms diverge easily. This paper proposes a novel nonparametric pose graph that models data association and SLAM in a single framework. An algorithm is further introduced to alternate between inferring data association and performing SLAM. Experimental results show that our approach has the new capability of associating object detections and localizing objects at the same time, leading to significantly better performance on both the data association and SLAM problems than achieved by considering only one and ignoring imperfections in the other.

Multi-robot deployments have the potential for completing tasks more efficiently. For example, in simultaneous localization and mapping (SLAM), robots can better localize themselves and the map if they can share measurements of each other (direct encounters) and of commonly observed parts of the map (indirect encounters). However, performance is contingent on the quality of the communications channel. In the underwater scenario, communicating over any appreciable distance is achieved using acoustics which is low-bandwidth, slow, and unreliable, making cooperative operations very challenging. In this paper, we present a framework for cooperative SLAM (C-SLAM) for multiple autonomous underwater vehicles (AUVs) communicating only through acoustics. We develop a novel graph-based C-SLAM algorithm that is able to (optimally) generate communication packets whose size scales linearly with the number of observed features since the last successful transmission, constantly with the number of vehicles in the collective, and does not grow with time even the case of dropped packets, which are common. As a result, AUVs can bound their localization error without the need for pre-installed beacons or surfacing for GPS fixes during navigation, leading to significant reduction in time required to complete missions. The proposed algorithm is validated through realistic marine vehicle and acoustic communication simulations.

Navigating in a previously unknown environment and recognizing naturally occurring text in a scene are two important autonomous capabilities that are typically treated as distinct. However, these two tasks are potentially complementary, (i) scene and pose priors can benefit text spotting, and (ii) the ability to identify and associate text features can benefit navigation accuracy through loop closures. Previous approaches to autonomous text spotting typically require significant training data and are too slow for real-time implementation. In this work, we propose a novel high-level feature descriptor, the “junction”, which is particularly well-suited to text representation and is also fast to compute. We show that we are able to improve SLAM through text spotting on datasets collected with a Google Tango, illustrating how location priors enable improved loop closure with text features.

This paper presents an overview of our research on accounting for robot pose uncertainty in area coverage applications. In the vast majority of existing literature on robotics area coverage, the location uncertainty of the robot is not considered. An uncertain robot pose results in an uncertain sensor swath, which in turn creates uncertainty about the achieved coverage. Here, we present a general framework where pose estimates are mapped through the coverage sensor model to obtain a probability of coverage over the discretized workspace. This probabilistic representation can then be used to adaptively plan paths for coverage based on an entropy reduction formulation. This framework is particularly well-suited to autonomous underwater vehicles (AUVs) performing seabed surveying operations. The AUV position estimate diverges from the actual AUV position while submerged due to the lack of a global position reference. This discrepancy can result in parts of the seabed being missed, which is unacceptable in safety-critical missions such as mine countermeasures. The proposed information-based path planning approach is able to guarantee area coverage even in the case of severe AUV position estimate drift. In-water experiments with an AUV show the effectiveness of the method.

Autonomous agents that can communicate and make relative measurements of each other can improve their collective localization accuracies. This is referred to as cooperative localization (CL). Autonomous underwater vehicle (AUV) CL is constrained by the low throughput, high latency, and unreliability of of the acoustic channel used to communicate when submerged. Here we propose a CL algorithm specifically designed for full trajectory, or maximum a posteriori, estimation for AUVs. The method is exact and has the advantage that the broadcast packet sizes increase only linearly with the number of AUVs in the collective and do not grow at all in the case of packet loss. The approach allows for AUV missions to be achieved more efficiently since: 1) vehicles waste less time surfacing for GPS fixes, and 2) payload data is more accurately localized through the smoothing approach.

Area coverage with an onboard sensor is an important task for an unmanned aerial vehicle (UAV) with many applications. Autonomous fixed-wing UAVs are more appropriate for larger scale area surveying since they can cover ground more quickly. However, their non-holonomic dynamics and susceptibility to disturbances make sensor coverage a challenging task. Most previous approaches to area coverage planning are offline and assume that the UAV can follow the planned trajectory exactly. In this paper, this restriction is removed as the aircraft maintains a coverage map based on its actual pose trajectory and makes control decisions based on that map. The aircraft is able to plan paths in situ based on sensor data and an accurate model of the on-board camera used for coverage. An information theoretic approach is used that selects desired headings that maximize the expected information gain over the coverage map. In addition, the branch entropy concept previously developed for autonomous underwater vehicles is extended to UAVs and ensures that the vehicle is able to achieve its global coverage mission. The coverage map over the workspace uses the projective camera model and compares the expected area of the target on the ground and the actual area covered on the ground by each pixel in the image. The camera is mounted on a two-axis gimbal and can either be stabilized or optimized for maximal coverage. Hardware-in-the-loop simulation results and real hardware implementation on a fixed-wing UAV show the effectiveness of the approach. By including the already developed automatic takeoff and landing capabilities, we now have a fully automated and robust platform for performing aerial imagery surveys.

Simultaneous Localization and Mapping, or SLAM, is required for mobile robots to be able to explore prior unknown space without a global positioning reference. While multiple robots can achieve the exploration task more quickly, this benefit comes with the cost of added complexity. Probabilistic occupancy grid maps from multiple agents must be merged in real-time without any prior knowledge of their relative transformation. In addition, the probabilistic information of the maps must be accounted for and fused accordingly. In this paper, a probabilistic version of the Generalized Voronoi Diagram (GVD), called the PGVD, is used to determine the relative transformation between maps and fuse them. The new method is effective for finding relative transformations quickly and reliably. In addition, the novel approach accounts for all map uncertainties in the fusion process.

One of the major problems for multi-robot SLAM is that the robots only know their positions in their own local coordinate frames, so fusing map data can be challenging. In this research, the mapping process is extended to multiple robots with a novel occupancy grid map fusion algorithm. Map fusion is achieved by transforming individual maps into the Hough space where they are represented in an abstract form. Properties of the Hough transform are used to find the common regions in the maps, which are then used to calculate the unknown transformation between the maps. Results are shown from tests performed on benchmark data sets and real-world experiments with multiple robotic platforms.

At present, autonomous underwater vehicle (AUV) mine countermeasure (MCM) surveys are pre-planned by operators using ladder or zig-zag paths. Such surveys are often conducted with side-looking sonar sensors whose performance is dependant on a number of environment factors, as well as lateral range from the AUV track. This research presents a sensor driven online approach to seabed coverage for MCM. A method is presented where paths are planned adaptively using a multi-objective optimization. Information theory is combined with a new concept coined branch entropy based on a hexagonal cell decomposition. The result is a planning algorithm that often produces shorter paths than conventional means and is also capable of accounting for environmental factors detected in situ. Hardware-in-the-loop simulations and in water trials conducted on the IVER2 AUV show the effectiveness of the proposed method.

The IEEE RAS Ontologies for Robotics and Automation Working Group is dedicated to developing a methodology for knowledge representation and reasoning in robotics and automation. As part of this working group, the Autonomous Robots sub-group is tasked with developing ontology modules for autonomous robots. This paper describes the work in progress on the development of ontologies for autonomous systems. For autonomous systems, the focus is on the cooperation, coordination, and communication of multiple unmanned aerial vehicles (UAVs), unmanned ground vehicles (UGVs), and autonomous underwater vehicles (AUVs). The ontologies serve as a framework for working out concepts of employment with multiple vehicles for a variety of operational scenarios with emphasis on collaborative and cooperative missions.

In this research, a decentralized platform for SLAM with multiple robots has been developed. An EKF-based single-robot SLAM is extended to multiple-robot SLAM with a novel occupancy grid map fusion algorithm. Map fusion is achieved through a multi-step process that includes image preprocessing, segmentation, cross correlation, approximating the relative transformation matrix, tuning of the transformation through the Radon image transform and similarity index, and then verification of the result using either map entropy or a verification index. Results are shown from tests performed in a real environment with multiple robotic platforms.

In this paper, a decentralized platform for Simultaneous Localization and Mapping (SLAM) with multiple robots is developed. A novel occupancy grid map fusion algorithm is proposed. Map fusion is achieved through a multi-step process that includes image pre-processing, map learning, relative transformation extraction and then verification of the results. The proposed map learning method is a process based on the Self Organizing Map (SOM). In the learning phase, the obstacles of the map are learned by clustering the occupied cells of the map. The clusters represent the spatial form of the map and make further analyses of the map easier and faster. Also, clusters can be interpreted as features extracted from the occupancy grid map so the map fusion problem becomes a task of matching features. Results of the experiments from tests performed on a real environment with multiple robots prove the effectiveness of the proposed solution.

In this paper, a decentralized platform for simultaneous localization and mapping (SLAM) with multiple robots is developed. Each robot performs single robot view-based SLAM using an extended Kalman filter to fuse data from two encoders and a laser ranger. To extend this approach to multiple robot SLAM, a novel occupancy grid map fusion algorithm is proposed. Map fusion is achieved through a multistep process that includes image preprocessing, map learning (clustering) using neural networks, relative orientation extraction using norm histogram cross correlation and a Radon transform, relative translation extraction using matching norm vectors, and then verification of the results. The proposed map learning method is a process based on the self-organizing map. In the learning phase, the obstacles of the map are learned by clustering the occupied cells of the map into clusters. The learning is an unsupervised process which can be done on the fly without any need to have output training patterns. The clusters represent the spatial form of the map and make further analyses of the map easier and faster. Also, clusters can be interpreted as features extracted from the occupancy grid map so the map fusion problem becomes a task of matching features. Results of the experiments from tests performed on a real environment with multiple robots prove the effectiveness of the proposed solution.