Katherine Driggs-Campbell

Yixuan Wang, Mingtong Zhang, Zhuoran Li et al. · mit, stanford

Scene representation is a crucial design choice in robotic manipulation systems. An ideal representation is expected to be 3D, dynamic, and semantic to meet the demands of diverse manipulation tasks. However, previous works often lack all three properties simultaneously. In this work, we introduce D$^3$Fields -- dynamic 3D descriptor fields. These fields are implicit 3D representations that take in 3D points and output semantic features and instance masks. They can also capture the dynamics of the underlying 3D environments. Specifically, we project arbitrary 3D points in the workspace onto multi-view 2D visual observations and interpolate features derived from visual foundational models. The resulting fused descriptor fields allow for flexible goal specifications using 2D images with varied contexts, styles, and instances. To evaluate the effectiveness of these descriptor fields, we apply our representation to rearrangement tasks in a zero-shot manner. Through extensive evaluation in real worlds and simulations, we demonstrate that D$^3$Fields are effective for zero-shot generalizable rearrangement tasks. We also compare D$^3$Fields with state-of-the-art implicit 3D representations and show significant improvements in effectiveness and efficiency.

Yixuan Wang, Yunzhu Li, Katherine Driggs-Campbell et al. · mit, stanford

Dynamics models learned from visual observations have shown to be effective in various robotic manipulation tasks. One of the key questions for learning such dynamics models is what scene representation to use. Prior works typically assume representation at a fixed dimension or resolution, which may be inefficient for simple tasks and ineffective for more complicated tasks. In this work, we investigate how to learn dynamic and adaptive representations at different levels of abstraction to achieve the optimal trade-off between efficiency and effectiveness. Specifically, we construct dynamic-resolution particle representations of the environment and learn a unified dynamics model using graph neural networks (GNNs) that allows continuous selection of the abstraction level. During test time, the agent can adaptively determine the optimal resolution at each model-predictive control (MPC) step. We evaluate our method in object pile manipulation, a task we commonly encounter in cooking, agriculture, manufacturing, and pharmaceutical applications. Through comprehensive evaluations both in the simulation and the real world, we show that our method achieves significantly better performance than state-of-the-art fixed-resolution baselines at the gathering, sorting, and redistribution of granular object piles made with various instances like coffee beans, almonds, corn, etc.

Tianchen Ji, Arun Narenthiran Sivakumar, Girish Chowdhary et al.

Despite the rapid advancement of navigation algorithms, mobile robots often produce anomalous behaviors that can lead to navigation failures. The ability to detect such anomalous behaviors is a key component in modern robots to achieve high-levels of autonomy. Reactive anomaly detection methods identify anomalous task executions based on the current robot state and thus lack the ability to alert the robot before an actual failure occurs. Such an alert delay is undesirable due to the potential damage to both the robot and the surrounding objects. We propose a proactive anomaly detection network (PAAD) for robot navigation in unstructured and uncertain environments. PAAD predicts the probability of future failure based on the planned motions from the predictive controller and the current observation from the perception module. Multi-sensor signals are fused effectively to provide robust anomaly detection in the presence of sensor occlusion as seen in field environments. Our experiments on field robot data demonstrates superior failure identification performance than previous methods, and that our model can capture anomalous behaviors in real-time while maintaining a low false detection rate in cluttered fields. Code, dataset, and video are available at https://github.com/tianchenji/PAAD

Shuijing Liu, Aamir Hasan, Kaiwen Hong et al.

Persons with visual impairments (PwVI) have difficulties understanding and navigating spaces around them. Current wayfinding technologies either focus solely on navigation or provide limited communication about the environment. Motivated by recent advances in visual-language grounding and semantic navigation, we propose DRAGON, a guiding robot powered by a dialogue system and the ability to associate the environment with natural language. By understanding the commands from the user, DRAGON is able to guide the user to the desired landmarks on the map, describe the environment, and answer questions from visual observations. Through effective utilization of dialogue, the robot can ground the user's free-form descriptions to landmarks in the environment, and give the user semantic information through spoken language. We conduct a user study with blindfolded participants in an everyday indoor environment. Our results demonstrate that DRAGON is able to communicate with the user smoothly, provide a good guiding experience, and connect users with their surrounding environment in an intuitive manner. Videos and code are available at https://sites.google.com/view/dragon-wayfinding/home.

Shuijing Liu, Peixin Chang, Zhe Huang et al.

We study the problem of safe and intention-aware robot navigation in dense and interactive crowds. Most previous reinforcement learning (RL) based methods fail to consider different types of interactions among all agents or ignore the intentions of people, which results in performance degradation. To learn a safe and efficient robot policy, we propose a novel recurrent graph neural network with attention mechanisms to capture heterogeneous interactions among agents through space and time. To encourage longsighted robot behaviors, we infer the intentions of dynamic agents by predicting their future trajectories for several timesteps. The predictions are incorporated into a model-free RL framework to prevent the robot from intruding into the intended paths of other agents. We demonstrate that our method enables the robot to achieve good navigation performance and non-invasiveness in challenging crowd navigation scenarios. We successfully transfer the policy learned in simulation to a real-world TurtleBot 2i. Our code and videos are available at https://sites.google.com/view/intention-aware-crowdnav/home.

Ye-Ji Mun, Masha Itkina, Shuijing Liu et al.

Autonomous navigation in crowded spaces poses a challenge for mobile robots due to the highly dynamic, partially observable environment. Occlusions are highly prevalent in such settings due to a limited sensor field of view and obstructing human agents. Previous work has shown that observed interactive behaviors of human agents can be used to estimate potential obstacles despite occlusions. We propose integrating such social inference techniques into the planning pipeline. We use a variational autoencoder with a specially designed loss function to learn representations that are meaningful for occlusion inference. This work adopts a deep reinforcement learning approach to incorporate the learned representation for occlusion-aware planning. In simulation, our occlusion-aware policy achieves comparable collision avoidance performance to fully observable navigation by estimating agents in occluded spaces. We demonstrate successful policy transfer from simulation to the real-world Turtlebot 2i. To the best of our knowledge, this work is the first to use social occlusion inference for crowd navigation.

Pulkit Katdare, Nan Jiang, Katherine Driggs-Campbell

Reinforcement Learning (RL) methods are typically sample-inefficient, making it challenging to train and deploy RL-policies in real world robots. Even a robust policy trained in simulation requires a real-world deployment to assess their performance. This paper proposes a new approach to evaluate the real-world performance of agent policies prior to deploying them in the real world. Our approach incorporates a simulator along with real-world offline data to evaluate the performance of any policy using the framework of Marginalized Importance Sampling (MIS). Existing MIS methods face two challenges: (1) large density ratios that deviate from a reasonable range and (2) indirect supervision, where the ratio needs to be inferred indirectly, thus exacerbating estimation error. Our approach addresses these challenges by introducing the target policy's occupancy in the simulator as an intermediate variable and learning the density ratio as the product of two terms that can be learned separately. The first term is learned with direct supervision and the second term has a small magnitude, thus making it computationally efficient. We analyze the sample complexity as well as error propagation of our two step-procedure. Furthermore, we empirically evaluate our approach on Sim2Sim environments such as Cartpole, Reacher, and Half-Cheetah. Our results show that our method generalizes well across a variety of Sim2Sim gap, target policies and offline data collection policies. We also demonstrate the performance of our algorithm on a Sim2Real task of validating the performance of a 7 DoF robotic arm using offline data along with the Gazebo simulator.

Haoran Yuan, Weigang Yi, Zhenyu Zhang et al.

Video-Action Models (VAMs) have emerged as a promising framework for embodied intelligence, learning implicit world dynamics from raw video streams to produce temporally consistent action predictions. Although such models demonstrate strong performance on long-horizon tasks through visual reasoning, they remain limited in contact-rich scenarios where critical interaction states are only partially observable from vision alone. In particular, fine-grained force modulation and contact transitions are not reliably encoded in visual tokens, leading to unstable or imprecise behaviors. To bridge this gap, we introduce the Video-Tactile Action Model (VTAM), a multimodal world modeling framework that incorporates tactile perception as a complementary grounding signal. VTAM augments a pretrained video transformer with tactile streams via a lightweight modality transfer finetuning, enabling efficient cross-modal representation learning without tactile-language paired data or independent tactile pretraining. To stabilize multimodal fusion, we introduce a tactile regularization loss that enforces balanced cross-modal attention, preventing visual latent dominance in the action model. VTAM demonstrates superior performance in contact-rich manipulation, maintaining a robust success rate of 90 percent on average. In challenging scenarios such as potato chip pick-and-place requiring high-fidelity force awareness, VTAM outperforms the pi 0.5 baseline by 80 percent. Our findings demonstrate that integrating tactile feedback is essential for correcting visual estimation errors in world action models, providing a scalable approach to physically grounded embodied foundation models.

Aamir Hasan, Neeloy Chakraborty, Haonan Chen et al.

Intelligent driving systems can be used to mitigate congestion through simple actions, thus improving many socioeconomic factors such as commute time and gas costs. However, these systems assume precise control over autonomous vehicle fleets, and are hence limited in practice as they fail to account for uncertainty in human behavior. Piecewise Constant (PC) Policies address these issues by structurally modeling the likeness of human driving to reduce traffic congestion in dense scenarios to provide action advice to be followed by human drivers. However, PC policies assume that all drivers behave similarly. To this end, we develop a co-operative advisory system based on PC policies with a novel driver trait conditioned Personalized Residual Policy, PeRP. PeRP advises drivers to behave in ways that mitigate traffic congestion. We first infer the driver's intrinsic traits on how they follow instructions in an unsupervised manner with a variational autoencoder. Then, a policy conditioned on the inferred trait adapts the action of the PC policy to provide the driver with a personalized recommendation. Our system is trained in simulation with novel driver modeling of instruction adherence. We show that our approach successfully mitigates congestion while adapting to different driver behaviors, with 4 to 22% improvement in average speed over baselines.

Pengcheng Wang, Kaiwen Hong, Chensheng Peng et al.

Unlike chatbots, physical AI must act while the world keeps evolving. Therefore, the inter-chunk pause of synchronous executors are fatal for dynamic tasks regardless of how fast the inference is. Asynchronous execution -- thinking while acting -- is therefore a structural requirement, and real-time chunking (RTC) makes it viable by recasting chunk transitions as inpainting: freezing committed actions and consistently generating the remainder. However, RTC with flow-matching policy is structurally suboptimal: its inpainting comes from inference-time corrections rather than the base policy, yielding little pre-training benefit, specific fine-tuning, heuristic guidance, and extra computation that inflates the latency. In this work, we observe that discrete diffusion policies, which generate actions by iteratively unmasking, are natural asynchronous executors that resolve all limitations at once: they are fine-tuning free since inpainting is their native operation, while early stopping further provides adaptive guidance and reduces inference cost. We propose DiscreteRTC, which replaces external corrections with native unmasking, and show on dynamic simulated benchmarks and real-world dynamic manipulation tasks that it achieves higher success rates than continuous RTC and other baselines. In summary, DiscreteRTC is simpler to implement with 0 lines of code for async inpainting, faster at inference with only 0.7x computation compared with generating actions from scratch, and better at execution with 50% higher success rate in real-world dynamic pick task compared with flow-matching-based RTC. More visualizations are on https://outsider86.github.io/DiscreteRTCSite/.

Peter Du, Katherine Driggs-Campbell

Uncovering potential failure cases is a crucial step in the validation of safety critical systems such as autonomous vehicles. Failure search may be done through logging substantial vehicle miles in either simulation or real world testing. Due to the sparsity of failure events, naive random search approaches require significant amounts of vehicle operation hours to find potential system weaknesses. As a result, adaptive searching techniques have been proposed to efficiently explore and uncover failure trajectories of an autonomous policy in simulation. Adaptive Stress Testing (AST) is one such method that poses the problem of failure search as a Markov decision process and uses reinforcement learning techniques to find high probability failures. However, this formulation requires a probability model for the actions of all agents in the environment. In systems where the environment actions are discrete and dependencies among agents exist, it may be infeasible to fully characterize the distribution or find a suitable proxy. This work proposes the use of a data driven approach to learn a suitable classifier that tries to model how humans identify {critical states and use this to guide failure search in AST. We show that the incorporation of critical states into the AST framework generates failure scenarios with increased safety violations in an autonomous driving policy with a discrete action space.

Aamir Hasan, Neeloy Chakraborty, Cathy Wu et al.

The effects of traffic congestion are widespread and are an impedance to everyday life. Piecewise constant driving policies have shown promise in helping mitigate traffic congestion in simulation environments. However, no works currently test these policies in situations involving real human users. Thus, we propose to evaluate these policies through the use of a shared control framework in a collaborative experiment with the human driver and the driving policy aiming to co-operatively mitigate congestion. We intend to use the CARLA simulator alongside the Flow framework to conduct user studies to evaluate the affect of piecewise constant driving policies. As such, we present our in-progress work in building our framework and discuss our proposed plan on evaluating this framework through a human-in-the-loop simulation user study.

Peter Du, Surya Murthy, Katherine Driggs-Campbell

As advances in artificial intelligence enable increasingly capable learning-based autonomous agents, it becomes more challenging for human observers to efficiently construct a mental model of the agent's behaviour. In order to successfully deploy autonomous agents, humans should not only be able to understand the individual limitations of the agents but also have insight on how they compare against one another. To do so, we need effective methods for generating human interpretable agent behaviour summaries. Single agent behaviour summarization has been tackled in the past through methods that generate explanations for why an agent chose to pick a particular action at a single timestep. However, for complex tasks, a per-action explanation may not be able to convey an agents global strategy. As a result, researchers have looked towards multi-timestep summaries which can better help humans assess an agents overall capability. More recently, multi-step summaries have also been used for generating contrasting examples to evaluate multiple agents. However, past approaches have largely relied on unstructured search methods to generate summaries and require agents to have a discrete action space. In this paper we present an adaptive search method for efficiently generating contrasting behaviour summaries with support for continuous state and action spaces. We perform a user study to evaluate the effectiveness of the summaries for helping humans discern the superior autonomous agent for a given task. Our results indicate that adaptive search can efficiently identify informative contrasting scenarios that enable humans to accurately select the better performing agent with a limited observation time budget.

Neeloy Chakraborty, Yixiao Fang, Andre Schreiber et al.

Teleoperation is an important technology to enable supervisors to control agricultural robots remotely. However, environmental factors in dense crop rows and limitations in network infrastructure hinder the reliability of data streamed to teleoperators. These issues result in delayed and variable frame rate video feeds that often deviate significantly from the robot's actual viewpoint. We propose a modular learning-based vision pipeline to generate delay-compensated images in real-time for supervisors. Our extensive offline evaluations demonstrate that our method generates more accurate images compared to state-of-the-art approaches in our setting. Additionally, ours is one of the few works to evaluate a delay-compensation method in outdoor field environments with complex terrain on data from a real robot in real-time. Resulting videos and code are provided at https://sites.google.com/illinois.edu/comp-teleop.

Fatemeh Cheraghi Pouria, Mahsa Golchoubian, Katherine Driggs-Campbell

Accurate trajectory prediction is critical for safe autonomous navigation in crowded environments. While many trajectory predictors output Gaussian distributions to represent the multi-modal distribution over future pedestrian positions, the reliability of their confidence levels often remains unaddressed. This limitation can lead to unsafe or overly conservative motion planning when the predictor is integrated with an uncertainty-aware planner. Existing Gaussian trajectory predictors primarily rely on the Negative Log-Likelihood loss, which is prone to predict over- or under-confident distributions, and may compromise downstream planner safety. This paper introduces a novel loss function for calibrating prediction uncertainty which leverages Kernel Density Estimation to estimate the empirical distribution of confidence levels. The proposed formulation enforces consistency with the properties of a Gaussian assumption by explicitly matching the estimated empirical distribution to the Chi-squared distribution. To ensure accurate mean prediction, a Mean Squared Error term is also incorporated in the final loss formulation. Experimental results on real-world trajectory datasets show that our method significantly improves the reliability of confidence levels predicted by different State-Of-The-Art Gaussian trajectory predictors. We also demonstrate the importance of providing planners with reliable probabilistic insights (i.e. calibrated confidence levels) for collision-free navigation in complex scenarios. For this purpose, we integrate Gaussian trajectory predictors trained with our loss function with an uncertainty-aware Model Predictive Control on scenarios extracted from real-world datasets, achieving improved planning performance through calibrated confidence levels.

Yuan Shen, Niviru Wijayaratne, Pranav Sriram et al.

The task of driver attention prediction has drawn considerable interest among researchers in robotics and the autonomous vehicle industry. Driver attention prediction can play an instrumental role in mitigating and preventing high-risk events, like collisions and casualties. However, existing driver attention prediction models neglect the distraction state and intention of the driver, which can significantly influence how they observe their surroundings. To address these issues, we present a new driver attention dataset, CoCAtt (Cognitive-Conditioned Attention). Unlike previous driver attention datasets, CoCAtt includes per-frame annotations that describe the distraction state and intention of the driver. In addition, the attention data in our dataset is captured in both manual and autopilot modes using eye-tracking devices of different resolutions. Our results demonstrate that incorporating the above two driver states into attention modeling can improve the performance of driver attention prediction. To the best of our knowledge, this work is the first to provide autopilot attention data. Furthermore, CoCAtt is currently the largest and the most diverse driver attention dataset in terms of autonomy levels, eye tracker resolutions, and driving scenarios. CoCAtt is available for download at https://cocatt-dataset.github.io.

Masha Itkina, Ye-Ji Mun, Katherine Driggs-Campbell et al.

Autonomous vehicles must reason about spatial occlusions in urban environments to ensure safety without being overly cautious. Prior work explored occlusion inference from observed social behaviors of road agents, hence treating people as sensors. Inferring occupancy from agent behaviors is an inherently multimodal problem; a driver may behave similarly for different occupancy patterns ahead of them (e.g., a driver may move at constant speed in traffic or on an open road). Past work, however, does not account for this multimodality, thus neglecting to model this source of aleatoric uncertainty in the relationship between driver behaviors and their environment. We propose an occlusion inference method that characterizes observed behaviors of human agents as sensor measurements, and fuses them with those from a standard sensor suite. To capture the aleatoric uncertainty, we train a conditional variational autoencoder with a discrete latent space to learn a multimodal mapping from observed driver trajectories to an occupancy grid representation of the view ahead of the driver. Our method handles multi-agent scenarios, combining measurements from multiple observed drivers using evidential theory to solve the sensor fusion problem. Our approach is validated on a cluttered, real-world intersection, outperforming baselines and demonstrating real-time capable performance. Our code is available at https://github.com/sisl/MultiAgentVariationalOcclusionInference .

Zhe Huang, Ruohua Li, Kazuki Shin et al.

Multi-pedestrian trajectory prediction is an indispensable element of autonomous systems that safely interact with crowds in unstructured environments. Many recent efforts in trajectory prediction algorithms have focused on understanding social norms behind pedestrian motions. Yet we observe these works usually hold two assumptions, which prevent them from being smoothly applied to robot applications: (1) positions of all pedestrians are consistently tracked, and (2) the target agent pays attention to all pedestrians in the scene. The first assumption leads to biased interaction modeling with incomplete pedestrian data. The second assumption introduces aggregation of redundant surrounding information, and the target agent may be affected by unimportant neighbors or present overly conservative motion. Thus, we propose Gumbel Social Transformer, in which an Edge Gumbel Selector samples a sparse interaction graph of partially detected pedestrians at each time step. A Node Transformer Encoder and a Masked LSTM encode pedestrian features with sampled sparse graphs to predict trajectories. We demonstrate that our model overcomes potential problems caused by the aforementioned assumptions, and our approach outperforms related works in trajectory prediction benchmarks. Code is available at \url{https://github.com/tedhuang96/gst}.

Zhe Huang, Aamir Hasan, Kazuki Shin et al.

Trajectory prediction is one of the key capabilities for robots to safely navigate and interact with pedestrians. Critical insights from human intention and behavioral patterns need to be integrated to effectively forecast long-term pedestrian behavior. Thus, we propose a framework incorporating a Mutable Intention Filter and a Warp LSTM (MIF-WLSTM) to simultaneously estimate human intention and perform trajectory prediction. The Mutable Intention Filter is inspired by particle filtering and genetic algorithms, where particles represent intention hypotheses that can be mutated throughout the pedestrian motion. Instead of predicting sequential displacement over time, our Warp LSTM learns to generate offsets on a full trajectory predicted by a nominal intention-aware linear model, which considers the intention hypotheses during filtering process. Through experiments on a publicly available dataset, we show that our method outperforms baseline approaches and demonstrate the robust performance of our method under abnormal intention-changing scenarios. Code is available at https://github.com/tedhuang96/mifwlstm.

Neeloy Chakraborty, Melkior Ornik, Katherine Driggs-Campbell

Autonomous systems are soon to be ubiquitous, spanning manufacturing, agriculture, healthcare, entertainment, and other industries. Most of these systems are developed with modular sub-components for decision-making, planning, and control that may be hand-engineered or learning-based. While these approaches perform well under the situations they were specifically designed for, they can perform especially poorly in out-of-distribution scenarios that will undoubtedly arise at test-time. The rise of foundation models trained on multiple tasks with impressively large datasets has led researchers to believe that these models may provide "common sense" reasoning that existing planners are missing, bridging the gap between algorithm development and deployment. While researchers have shown promising results in deploying foundation models to decision-making tasks, these models are known to hallucinate and generate decisions that may sound reasonable, but are in fact poor. We argue there is a need to step back and simultaneously design systems that can quantify the certainty of a model's decision, and detect when it may be hallucinating. In this work, we discuss the current use cases of foundation models for decision-making tasks, provide a general definition for hallucinations with examples, discuss existing approaches to hallucination detection and mitigation with a focus on decision problems, present guidelines, and explore areas for further research in this exciting field.

Shuijing Liu, Haochen Xia, Fatemeh Cheraghi Pouria et al.

We study the problem of robot navigation in dense and interactive crowds with environmental constraints such as corridors and furniture. Previous methods fail to consider all types of interactions among agents and obstacles, leading to unsafe and inefficient robot paths. In this article, we leverage a graph-based representation of crowded and constrained scenarios and propose a structured framework to learn robot navigation policies with deep reinforcement learning. We first split the representations of different components in the environment and propose a heterogeneous spatio-temporal (st) graph to model distinct interactions among humans, robots, and obstacles. Based on the heterogeneous st-graph, we propose HEIGHT, a novel navigation policy network architecture with different components to capture heterogeneous interactions among entities through space and time. HEIGHT utilizes attention mechanisms to prioritize important interactions and a recurrent network to track changes in the dynamic scene over time, encouraging the robot to avoid collisions adaptively. Through extensive simulation and real-world experiments, we demonstrate that HEIGHT outperforms state-of-the-art baselines in terms of success and efficiency in challenging navigation scenarios. Furthermore, we demonstrate that our pipeline achieves better zero-shot generalization capability than previous works when the densities of humans and obstacles change. More videos are available at https://sites.google.com/view/crowdnav-height/home.

Haonan Chen, Jiaming Xu, Lily Sheng et al.

When performing tasks like laundry, humans naturally coordinate both hands to manipulate objects and anticipate how their actions will change the state of the clothes. However, achieving such coordination in robotics remains challenging due to the need to model object movement, predict future states, and generate precise bimanual actions. In this work, we address these challenges by infusing the predictive nature of human manipulation strategies into robot imitation learning. Specifically, we disentangle task-related state transitions from agent-specific inverse dynamics modeling to enable effective bimanual coordination. Using a demonstration dataset, we train a diffusion model to predict future states given historical observations, envisioning how the scene evolves. Then, we use an inverse dynamics model to compute robot actions that achieve the predicted states. Our key insight is that modeling object movement can help learning policies for bimanual coordination manipulation tasks. Evaluating our framework across diverse simulation and real-world manipulation setups, including multimodal goal configurations, bimanual manipulation, deformable objects, and multi-object setups, we find that it consistently outperforms state-of-the-art state-to-action mapping policies. Our method demonstrates a remarkable capacity to navigate multimodal goal configurations and action distributions, maintain stability across different control modes, and synthesize a broader range of behaviors than those present in the demonstration dataset.

Tianchen Ji, Neeloy Chakraborty, Andre Schreiber et al.

As automated vehicles enter public roads, safety in a near-infinite number of driving scenarios becomes one of the major concerns for the widespread adoption of fully autonomous driving. The ability to detect anomalous situations outside of the operational design domain is a key component in self-driving cars, enabling us to mitigate the impact of abnormal ego behaviors and to realize trustworthy driving systems. On-road anomaly detection in egocentric videos remains a challenging problem due to the difficulties introduced by complex and interactive scenarios. We conduct a holistic analysis of common on-road anomaly patterns, from which we propose three unsupervised anomaly detection experts: a scene expert that focuses on frame-level appearances to detect abnormal scenes and unexpected scene motions; an interaction expert that models normal relative motions between two road participants and raises alarms whenever anomalous interactions emerge; and a behavior expert which monitors abnormal behaviors of individual objects by future trajectory prediction. To combine the strengths of all the modules, we propose an expert ensemble (Xen) using a Kalman filter, in which the final anomaly score is absorbed as one of the states and the observations are generated by the experts. Our experiments employ a novel evaluation protocol for realistic model performance, demonstrate superior anomaly detection performance than previous methods, and show that our framework has potential in classifying anomaly types using unsupervised learning on a large-scale on-road anomaly dataset.

Haonan Chen, Cheng Zhu, Shuijing Liu et al.

Tool use is essential for enabling robots to perform complex real-world tasks, but learning such skills requires extensive datasets. While teleoperation is widely used, it is slow, delay-sensitive, and poorly suited for dynamic tasks. In contrast, human videos provide a natural way for data collection without specialized hardware, though they pose challenges on robot learning due to viewpoint variations and embodiment gaps. To address these challenges, we propose a framework that transfers tool-use knowledge from humans to robots. To improve the policy's robustness to viewpoint variations, we use two RGB cameras to reconstruct 3D scenes and apply Gaussian splatting for novel view synthesis. We reduce the embodiment gap using segmented observations and tool-centric, task-space actions to achieve embodiment-invariant visuomotor policy learning. We demonstrate our framework's effectiveness across a diverse suite of tool-use tasks, where our learned policy shows strong generalization and robustness to human perturbations, camera motion, and robot base movement. Our method achieves a 71\% improvement in task success over teleoperation-based diffusion policies and dramatically reduces data collection time by 77\% and 41\% compared to teleoperation and the state-of-the-art interface, respectively.

Haonan Chen, Jiaming Xu, Hongyu Chen et al.

Effectively integrating diverse sensory modalities is crucial for robotic manipulation. However, the typical approach of feature concatenation is often suboptimal: dominant modalities such as vision can overwhelm sparse but critical signals like touch in contact-rich tasks, and monolithic architectures cannot flexibly incorporate new or missing modalities without retraining. Our method factorizes the policy into a set of diffusion models, each specialized for a single representation (e.g., vision or touch), and employs a router network that learns consensus weights to adaptively combine their contributions, enabling incremental of new representations. We evaluate our approach on simulated manipulation tasks in {RLBench}, as well as real-world tasks such as occluded object picking, in-hand spoon reorientation, and puzzle insertion, where it significantly outperforms feature-concatenation baselines on scenarios requiring multimodal reasoning. Our policy further demonstrates robustness to physical perturbations and sensor corruption. We further conduct perturbation-based importance analysis, which reveals adaptive shifts between modalities.

Neeloy Chakraborty, John Pohovey, Melkior Ornik et al.

Large language models (LLMs) have recently demonstrated success in generalizing across decision-making tasks including planning, control, and prediction, but their tendency to hallucinate unsafe and undesired outputs poses risks. We argue that detecting such failures is necessary, especially in safety-critical scenarios. Existing methods for black-box models often detect hallucinations by identifying inconsistencies across multiple samples. Many of these approaches typically introduce prompt perturbations like randomizing detail order or generating adversarial inputs, with the intuition that a confident model should produce stable outputs. We first perform a manual case study showing that other forms of perturbations (e.g., adding noise, removing sensor details) cause LLMs to hallucinate in a multi-agent driving environment. We then propose a novel method for efficiently searching the space of prompt perturbations using adaptive stress testing (AST) with Monte-Carlo tree search (MCTS). Our AST formulation enables discovery of scenarios and prompts that cause language models to act with high uncertainty or even crash. By generating MCTS prompt perturbation trees across diverse scenarios, we show through extensive experiments that offline analyses can be used at runtime to automatically generate prompts that influence model uncertainty, and to inform real-time trust assessments of an LLM. We further characterize LLMs deployed as planners in a single-agent lunar lander environment and in a multi-agent robot crowd navigation simulation. Overall, ours is one of the first hallucination intervention algorithms to pave a path towards rigorous characterization of black-box LLM planners.

Aamir Hasan, Neeloy Chakraborty, Haonan Chen et al.

Fleets of autonomous vehicles can mitigate traffic congestion through simple actions, thus improving many socioeconomic factors such as commute time and gas costs. However, these approaches are limited in practice as they assume precise control over autonomous vehicle fleets, incur extensive installation costs for a centralized sensor ecosystem, and also fail to account for uncertainty in driver behavior. To this end, we develop a class of learned residual policies that can be used in cooperative advisory systems and only require the use of a single vehicle with a human driver. Our policies advise drivers to behave in ways that mitigate traffic congestion while accounting for diverse driver behaviors, particularly drivers' reactions to instructions, to provide an improved user experience. To realize such policies, we introduce an improved reward function that explicitly addresses congestion mitigation and driver attitudes to advice. We show that our residual policies can be personalized by conditioning them on an inferred driver trait that is learned in an unsupervised manner with a variational autoencoder. Our policies are trained in simulation with our novel instruction adherence driver model, and evaluated in simulation and through a user study (N=16) to capture the sentiments of human drivers. Our results show that our approaches successfully mitigate congestion while adapting to different driver behaviors, with up to 20% and 40% improvement as measured by a combination metric of speed and deviations in speed across time over baselines in our simulation tests and user study, respectively. Our user study further shows that our policies are human-compatible and personalize to drivers.

Aamir Hasan, Neeloy Chakraborty, Haonan Chen et al.

Real-time Advisory (RTA) systems, such as navigational and eco-driving assistants, are becoming increasingly ubiquitous in vehicles due to their benefits for users and society. Until autonomous vehicles mature, such advisory systems will continue to expand their ability to cooperate with drivers, enabling safer and more eco-friendly driving practices while improving user experience. However, the interactions between these systems and drivers have not been studied extensively. To this end, we conduct a driving simulator study (N=16) to capture driver reactions to a Cooperative RTA system. Through a case study with a congestion mitigation assistant, we qualitatively analyze the sentiments of drivers towards advisory systems and discuss driver preferences for various aspects of the interaction. We comment on how the advice should be communicated, the effects of the advice on driver trust, and how drivers adapt to the system. We present recommendations to inform the future design of Cooperative RTA systems.

Zhe Huang, John Pohovey, Ananya Yammanuru et al.

Large Language Models (LLM) and Vision Language Models (VLM) enable robots to ground natural language prompts into control actions to achieve tasks in an open world. However, when applied to a long-horizon collaborative task, this formulation results in excessive prompting for initiating or clarifying robot actions at every step of the task. We propose Language-driven Intention Tracking (LIT), leveraging LLMs and VLMs to model the human user's long-term behavior and to predict the next human intention to guide the robot for proactive collaboration. We demonstrate smooth coordination between a LIT-based collaborative robot and the human user in collaborative cooking tasks.

Abhi Kamboj, Katherine Driggs-Campbell

Robotic grasping presents a difficult motor task in real-world scenarios, constituting a major hurdle to the deployment of capable robots across various industries. Notably, the scarcity of data makes grasping particularly challenging for learned models. Recent advancements in computer vision have witnessed a growth of successful unsupervised training mechanisms predicated on massive amounts of data sourced from the Internet, and now nearly all prominent models leverage pretrained backbone networks. Against this backdrop, we begin to investigate the potential benefits of large-scale visual pretraining in enhancing robot grasping performance. This preliminary literature review sheds light on critical challenges and delineates prospective directions for future research in visual pretraining for robotic manipulation.

Pulkit Katdare, Anant Joshi, Katherine Driggs-Campbell

Policy gradient methods are a vital ingredient behind the success of modern reinforcement learning. Modern policy gradient methods, although successful, introduce a residual error in gradient estimation. In this work, we argue that this residual term is significant and correcting for it could potentially improve sample-complexity of reinforcement learning methods. To that end, we propose log density gradient to estimate the policy gradient, which corrects for this residual error term. Log density gradient method computes policy gradient by utilising the state-action discounted distributional formulation. We first present the equations needed to exactly find the log density gradient for a tabular Markov Decision Processes (MDPs). For more complex environments, we propose a temporal difference (TD) method that approximates log density gradient by utilizing backward on-policy samples. Since backward sampling from a Markov chain is highly restrictive we also propose a min-max optimization that can approximate log density gradient using just on-policy samples. We also prove uniqueness, and convergence under linear function approximation, for this min-max optimization. Finally, we show that the sample complexity of our min-max optimization to be of the order of $m^{-1/2}$, where $m$ is the number of on-policy samples. We also demonstrate a proof-of-concept for our log density gradient method on gridworld environment, and observe that our method is able to improve upon the classical policy gradient method by a clear margin, thus indicating a promising novel direction to develop reinforcement learning algorithms that require fewer samples.

Sourya Basu, Pulkit Katdare, Prasanna Sattigeri et al.

Efficient transfer learning algorithms are key to the success of foundation models on diverse downstream tasks even with limited data. Recent works of Basu et al. (2023) and Kaba et al. (2022) propose group averaging (equitune) and optimization-based methods, respectively, over features from group-transformed inputs to obtain equivariant outputs from non-equivariant neural networks. While Kaba et al. (2022) are only concerned with training from scratch, we find that equitune performs poorly on equivariant zero-shot tasks despite good finetuning results. We hypothesize that this is because pretrained models provide better quality features for certain transformations than others and simply averaging them is deleterious. Hence, we propose λ-equitune that averages the features using importance weights, λs. These weights are learned directly from the data using a small neural network, leading to excellent zero-shot and finetuned results that outperform equitune. Further, we prove that λ-equitune is equivariant and a universal approximator of equivariant functions. Additionally, we show that the method of Kaba et al. (2022) used with appropriate loss functions, which we call equizero, also gives excellent zero-shot and finetuned performance. Both equitune and equizero are special cases of λ-equitune. To show the simplicity and generality of our method, we validate on a wide range of diverse applications and models such as 1) image classification using CLIP, 2) deep Q-learning, 3) fairness in natural language generation (NLG), 4) compositional generalization in languages, and 5) image classification using pretrained CNNs such as Resnet and Alexnet.

Aamir Hasan, Pranav Sriram, Katherine Driggs-Campbell

Spatio-temporal graphs (ST-graphs) have been used to model time series tasks such as traffic forecasting, human motion modeling, and action recognition. The high-level structure and corresponding features from ST-graphs have led to improved performance over traditional architectures. However, current methods tend to be limited by simple features, despite the rich information provided by the full graph structure, which leads to inefficiencies and suboptimal performance in downstream tasks. We propose the use of features derived from meta-paths, walks across different types of edges, in ST-graphs to improve the performance of Structural Recurrent Neural Network. In this paper, we present the Meta-path Enhanced Structural Recurrent Neural Network (MESRNN), a generic framework that can be applied to any spatio-temporal task in a simple and scalable manner. We employ MESRNN for pedestrian trajectory prediction, utilizing these meta-path based features to capture the relationships between the trajectories of pedestrians at different points in time and space. We compare our MESRNN against state-of-the-art ST-graph methods on standard datasets to show the performance boost provided by meta-path information. The proposed model consistently outperforms the baselines in trajectory prediction over long time horizons by over 32\%, and produces more socially compliant trajectories in dense crowds. For more information please refer to the project website at https://sites.google.com/illinois.edu/mesrnn/home.

Pulkit Katdare, Shuijing Liu, Katherine Driggs-Campbell

Applying reinforcement learning (RL) methods on robots typically involves training a policy in simulation and deploying it on a robot in the real world. Because of the model mismatch between the real world and the simulator, RL agents deployed in this manner tend to perform suboptimally. To tackle this problem, researchers have developed robust policy learning algorithms that rely on synthetic noise disturbances. However, such methods do not guarantee performance in the target environment. We propose a convex risk minimization algorithm to estimate the model mismatch between the simulator and the target domain using trajectory data from both environments. We show that this estimator can be used along with the simulator to evaluate performance of an RL agents in the target domain, effectively bridging the gap between these two environments. We also show that the convergence rate of our estimator to be of the order of ${n^{-1/4}}$, where $n$ is the number of training samples. In simulation, we demonstrate how our method effectively approximates and evaluates performance on Gridworld, Cartpole, and Reacher environments on a range of policies. We also show that the our method is able to estimate performance of a 7 DOF robotic arm using the simulator and remotely collected data from the robot in the real world.

Shuijing Liu, Peixin Chang, Haonan Chen et al.

Navigation through uncontrolled intersections is one of the key challenges for autonomous vehicles. Identifying the subtle differences in hidden traits of other drivers can bring significant benefits when navigating in such environments. We propose an unsupervised method for inferring driver traits such as driving styles from observed vehicle trajectories. We use a variational autoencoder with recurrent neural networks to learn a latent representation of traits without any ground truth trait labels. Then, we use this trait representation to learn a policy for an autonomous vehicle to navigate through a T-intersection with deep reinforcement learning. Our pipeline enables the autonomous vehicle to adjust its actions when dealing with drivers of different traits to ensure safety and efficiency. Our method demonstrates promising performance and outperforms state-of-the-art baselines in the T-intersection scenario.

Peixin Chang, Shuijing Liu, Katherine Driggs-Campbell

Inspired by sensorimotor theory, we propose a novel pipeline for task-oriented voice-controlled robots. Previous method relies on a large amount of labels as well as task-specific reward functions. Not only can such an approach hardly be improved after the deployment, but also has limited generalization across robotic platforms and tasks. To address these problems, we learn a visual-audio representation (VAR) that associates images and sound commands with minimal supervision. Using this representation, we generate an intrinsic reward function to learn robot policies with reinforcement learning, which eliminates the laborious reward engineering process. We demonstrate our approach on various robotic platforms, where the robots hear an audio command, identify the associated target object, and perform precise control to fulfill the sound command. We show that our method outperforms previous work across various sound types and robotic tasks even with fewer amount of labels. We successfully deploy the policy learned in a simulator to a real Kinova Gen3. We also demonstrate that our VAR and the intrinsic reward function allows the robot to improve itself using only a small amount of labeled data collected in the real world.

Yuan Shen, Niviru Wijayaratne, Katherine Driggs-Campbell

Effective human-vehicle collaboration requires an appropriate un-derstanding of vehicle behavior for safety and trust. Improvingon our prior work by adding a future prediction module, we in-troduce our framework, calledAutoPreview, to enable humans topreview autopilot behaviors prior to direct interaction with thevehicle. Previewing autopilot behavior can help to ensure smoothhuman-vehicle collaboration during the initial exploration stagewith the vehicle. To demonstrate its practicality, we conducted acase study on human-vehicle collaboration and built a prototypeof our framework with the CARLA simulator. Additionally, weconducted a between-subject control experiment (n=10) to studywhether ourAutoPreviewframework can provide a deeper under-standing of autopilot behavior compared to direct interaction. Ourresults suggest that theAutoPreviewframework does, in fact, helpusers understand autopilot behavior and develop appropriate men-tal models

Tianchen Ji, Sri Theja Vuppala, Girish Chowdhary et al.

To achieve high-levels of autonomy, modern robots require the ability to detect and recover from anomalies and failures with minimal human supervision. Multi-modal sensor signals could provide more information for such anomaly detection tasks; however, the fusion of high-dimensional and heterogeneous sensor modalities remains a challenging problem. We propose a deep learning neural network: supervised variational autoencoder (SVAE), for failure identification in unstructured and uncertain environments. Our model leverages the representational power of VAE to extract robust features from high-dimensional inputs for supervised learning tasks. The training objective unifies the generative model and the discriminative model, thus making the learning a one-stage procedure. Our experiments on real field robot data demonstrate superior failure identification performance than baseline methods, and that our model learns interpretable representations. Videos of our results are available on our website: https://sites.google.com/illinois.edu/supervised-vae .

Shuijing Liu, Peixin Chang, Weihang Liang et al.

Safe and efficient navigation through human crowds is an essential capability for mobile robots. Previous work on robot crowd navigation assumes that the dynamics of all agents are known and well-defined. In addition, the performance of previous methods deteriorates in partially observable environments and environments with dense crowds. To tackle these problems, we propose decentralized structural-Recurrent Neural Network (DS-RNN), a novel network that reasons about spatial and temporal relationships for robot decision making in crowd navigation. We train our network with model-free deep reinforcement learning without any expert supervision. We demonstrate that our model outperforms previous methods in challenging crowd navigation scenarios. We successfully transfer the policy learned in the simulator to a real-world TurtleBot 2i. For more information, please visit the project website at https://sites.google.com/view/crowdnav-ds-rnn/home.

Raunak P. Bhattacharyya, Kyle Brown, Juanran Wang et al.

An open problem in autonomous driving research is modeling human driving behavior, which is needed for the planning component of the autonomy stack, safety validation through traffic simulation, and causal inference for generating explanations for autonomous driving. Modeling human driving behavior is challenging because it is stochastic, high-dimensional, and involves interaction between multiple agents. This problem has been studied in various communities with a vast body of literature. Existing reviews have generally focused on one aspect: motion prediction. In this article, we present a unification of the literature that covers intent estimation, trait estimation, and motion prediction. This unification is enabled by modeling multi-agent driving as a partially observable stochastic game, which allows us to cast driver modeling tasks as inference problems. We classify driver models into a taxonomy based on the specific tasks they address and the key attributes of their approach. Finally, we identify open research opportunities in the field of driver modeling.

Xiaobai Ma, Katherine Driggs-Campbell, Zongzhang Zhang et al.

Gradient-based methods are often used for policy optimization in deep reinforcement learning, despite being vulnerable to local optima and saddle points. Although gradient-free methods (e.g., genetic algorithms or evolution strategies) help mitigate these issues, poor initialization and local optima are still concerns in highly nonconvex spaces. This paper presents a method for policy optimization based on Monte-Carlo tree search and gradient-free optimization. Our method, called Monte-Carlo tree search for policy optimization (MCTSPO), provides a better exploration-exploitation trade-off through the use of the upper confidence bound heuristic. We demonstrate improved performance on reinforcement learning tasks with deceptive or sparse reward functions compared to popular gradient-based and deep genetic algorithm baselines.

Peter Du, Zhe Huang, Tianqi Liu et al.

As autonomous systems begin to operate amongst humans, methods for safe interaction must be investigated. We consider an example of a small autonomous vehicle in a pedestrian zone that must safely maneuver around people in a free-form fashion. We investigate two key questions: How can we effectively integrate pedestrian intent estimation into our autonomous stack. Can we develop an online monitoring framework to give formal guarantees on the safety of such human-robot interactions. We present a pedestrian intent estimation framework that can accurately predict future pedestrian trajectories given multiple possible goal locations. We integrate this into a reachability-based online monitoring scheme that formally assesses the safety of these interactions with nearly real-time performance (approximately 0.3 seconds). These techniques are integrated on a test vehicle with a complete in-house autonomous stack, demonstrating effective and safe interaction in real-world experiments.

Peixin Chang, Shuijing Liu, Haonan Chen et al.

We explore the interpretation of sound for robot decision making, inspired by human speech comprehension. While previous methods separate sound processing unit and robot controller, we propose an end-to-end deep neural network which directly interprets sound commands for visual-based decision making. The network is trained using reinforcement learning with auxiliary losses on the sight and sound networks. We demonstrate our approach on two robots, a TurtleBot3 and a Kuka-IIWA arm, which hear a command word, identify the associated target object, and perform precise control to reach the target. For both robots, we show the effectiveness of our network in generalization to sound types and robotic tasks empirically. We successfully transfer the policy learned in simulator to a real-world TurtleBot3.

Anthony Corso, Peter Du, Katherine Driggs-Campbell et al.

Determining possible failure scenarios is a critical step in the evaluation of autonomous vehicle systems. Real-world vehicle testing is commonly employed for autonomous vehicle validation, but the costs and time requirements are high. Consequently, simulation-driven methods such as Adaptive Stress Testing (AST) have been proposed to aid in validation. AST formulates the problem of finding the most likely failure scenarios as a Markov decision process, which can be solved using reinforcement learning. In practice, AST tends to find scenarios where failure is unavoidable and tends to repeatedly discover the same types of failures of a system. This work addresses these issues by encoding domain relevant information into the search procedure. With this modification, the AST method discovers a larger and more expressive subset of the failure space when compared to the original AST formulation. We show that our approach is able to identify useful failure scenarios of an autonomous vehicle policy.

Carl-Johan Hoel, Katherine Driggs-Campbell, Krister Wolff et al.

Tactical decision making for autonomous driving is challenging due to the diversity of environments, the uncertainty in the sensor information, and the complex interaction with other road users. This paper introduces a general framework for tactical decision making, which combines the concepts of planning and learning, in the form of Monte Carlo tree search and deep reinforcement learning. The method is based on the AlphaGo Zero algorithm, which is extended to a domain with a continuous state space where self-play cannot be used. The framework is applied to two different highway driving cases in a simulated environment and it is shown to perform better than a commonly used baseline method. The strength of combining planning and learning is also illustrated by a comparison to using the Monte Carlo tree search or the neural network policy separately.

Masha Itkina, Katherine Driggs-Campbell, Mykel J. Kochenderfer

A key challenge for autonomous driving is safe trajectory planning in cluttered, urban environments with dynamic obstacles, such as pedestrians, bicyclists, and other vehicles. A reliable prediction of the future environment, including the behavior of dynamic agents, would allow planning algorithms to proactively generate a trajectory in response to a rapidly changing environment. We present a novel framework that predicts the future occupancy state of the local environment surrounding an autonomous agent by learning a motion model from occupancy grid data using a neural network. We take advantage of the temporal structure of the grid data by utilizing a convolutional long-short term memory network in the form of the PredNet architecture. This method is validated on the KITTI dataset and demonstrates higher accuracy and better predictive power than baseline methods.

Raunak P. Bhattacharyya, Derek J. Phillips, Changliu Liu et al.

Recent developments in multi-agent imitation learning have shown promising results for modeling the behavior of human drivers. However, it is challenging to capture emergent traffic behaviors that are observed in real-world datasets. Such behaviors arise due to the many local interactions between agents that are not commonly accounted for in imitation learning. This paper proposes Reward Augmented Imitation Learning (RAIL), which integrates reward augmentation into the multi-agent imitation learning framework and allows the designer to specify prior knowledge in a principled fashion. We prove that convergence guarantees for the imitation learning process are preserved under the application of reward augmentation. This method is validated in a driving scenario, where an entire traffic scene is controlled by driving policies learned using our proposed algorithm. Further, we demonstrate improved performance in comparison to traditional imitation learning algorithms both in terms of the local actions of a single agent and the behavior of emergent properties in complex, multi-agent settings.

Xiaobai Ma, Katherine Driggs-Campbell, Mykel J. Kochenderfer

To improve efficiency and reduce failures in autonomous vehicles, research has focused on developing robust and safe learning methods that take into account disturbances in the environment. Existing literature in robust reinforcement learning poses the learning problem as a two player game between the autonomous system and disturbances. This paper examines two different algorithms to solve the game, Robust Adversarial Reinforcement Learning and Neural Fictitious Self Play, and compares performance on an autonomous driving scenario. We extend the game formulation to a semi-competitive setting and demonstrate that the resulting adversary better captures meaningful disturbances that lead to better overall performance. The resulting robust policy exhibits improved driving efficiency while effectively reducing collision rates compared to baseline control policies produced by traditional reinforcement learning methods.

Michael Kelly, Chelsea Sidrane, Katherine Driggs-Campbell et al.

Imitation learning has proven to be useful for many real-world problems, but approaches such as behavioral cloning suffer from data mismatch and compounding error issues. One attempt to address these limitations is the DAgger algorithm, which uses the state distribution induced by the novice to sample corrective actions from the expert. Such sampling schemes, however, require the expert to provide action labels without being fully in control of the system. This can decrease safety and, when using humans as experts, is likely to degrade the quality of the collected labels due to perceived actuator lag. In this work, we propose HG-DAgger, a variant of DAgger that is more suitable for interactive imitation learning from human experts in real-world systems. In addition to training a novice policy, HG-DAgger also learns a safety threshold for a model-uncertainty-based risk metric that can be used to predict the performance of the fully trained novice in different regions of the state space. We evaluate our method on both a simulated and real-world autonomous driving task, and demonstrate improved performance over both DAgger and behavioral cloning.

Kunal Menda, Katherine Driggs-Campbell, Mykel J. Kochenderfer

While imitation learning is often used in robotics, the approach frequently suffers from data mismatch and compounding errors. DAgger is an iterative algorithm that addresses these issues by aggregating training data from both the expert and novice policies, but does not consider the impact of safety. We present a probabilistic extension to DAgger, which attempts to quantify the confidence of the novice policy as a proxy for safety. Our method, EnsembleDAgger, approximates a Gaussian Process using an ensemble of neural networks. Using the variance as a measure of confidence, we compute a decision rule that captures how much we doubt the novice, thus determining when it is safe to allow the novice to act. With this approach, we aim to maximize the novice's share of actions, while constraining the probability of failure. We demonstrate improved safety and learning performance compared to other DAgger variants and classic imitation learning on an inverted pendulum and in the MuJoCo HalfCheetah environment.