We are the Control & Trustworthy Robotics Lab (CTRL) at the University of Washington, a research lab developing trustworthy autonomous systems that can operate seamlessly with, alongside, and around humans.
CTRL focuses on developing safe, intelligent, and trustworthy robotics systems that can operate seamlessly with, alongside, and around humans. Our work is at the intersection of control theory, machine learning, robotics, and formal methods. The ultimate goal is to achieve human trust in learning-enabled robot autonomy, starting from the algorithmic foundations of safe robot decision-making and control, and incorporating further refinement through learnings from practical deployment.
Research Areas
Below are research thrusts that CTRL is working on:
Safe Interaction-aware Planning and Control
The next wave of robotic systems will operate in dynamic, stochastic, and unstructured environments, especially settings where robots must interact with humans.
This line of research develops techniques to model human interaction dynamics and provide safety assurances throughout a potentially learning-enabled robot autonomy stack to enable intelligent yet safe operations.
Relevant works
Multimodal Probabilistic Model-Based Planning for Human-Robot Interaction
Schmerling, E.,
Leung, K.,
Vollprecht, W.,
and Pavone, M.
In Proc. IEEE Conf. on Robotics and Automation,
2018
This paper presents a method for constructing human-robot interaction policies in settings where multimodality, i.e., the possibility of multiple highly distinct futures, plays a critical role in decision making. We are motivated in this work by the example of traffic weaving, e.g., at highway onramps/offramps, where entering and exiting cars must swap lanes in a short distance — a challenging negotiation even for experienced drivers due to the inherent multimodal uncertainty of who will pass whom. Our approach is to learn multimodal probability distributions over future human actions from a dataset of human-human exemplars and perform real-time robot policy construction in the resulting environment model through massively parallel sampling of human responses to candidate robot action sequences. Direct learning of these distributions is made possible by recent advances in the theory of conditional variational autoencoders (CVAEs), whereby we learn action distributions simultaneously conditioned on the present interaction history, as well as candidate future robot actions in order to take into account response dynamics. We demonstrate the efficacy of this approach with a human-in-the-loop simulation of a traffic weaving scenario.
@inproceedings{SchmerlingLeungEtAl2018,author={Schmerling, E. and Leung, K. and Vollprecht, W. and Pavone, M.},booktitle={{Proc.\ IEEE Conf.\ on Robotics and Automation}},title={{Multimodal Probabilistic Model-Based Planning for Human-Robot Interaction}},year={2018},arxiv={1710.09483},category={interaction},img={SchmerlingLeungEtAl2018.png},selected={true}}
Multimodal Deep Generative Models for Trajectory Prediction: A Conditional Variational Autoencoder Approach
Ivanovic, B.*,
Leung, K.*,
Schmerling, E.,
and Pavone, M.
Human behavior prediction models enable robots to anticipate how humans may react to their actions, and hence are instrumental to devising safe and proactive robot planning algorithms. However, modeling complex interaction dynamics and capturing the possibility of many possible outcomes in such interactive settings is very challenging, which has recently prompted the study of several different approaches. In this work, we provide a self-contained tutorial on a conditional variational autoencoder (CVAE) approach to human behavior prediction which, at its core, can produce a multimodal probability distribution over future human trajectories conditioned on past interactions and candidate robot future actions. Specifically, the goals of this tutorial paper are to review and build a taxonomy of state-of-the-art methods in human behavior prediction, from physics-based to purely data-driven methods, provide a rigorous yet easily accessible description of a data-driven, CVAE-based approach, highlight important design characteristics that make this an attractive model to use in the context of model-based planning for human-robot interactions, and provide important design considerations when using this class of models.
@article{IvanovicLeungEtAl2020,author={Ivanovic, B.* and Leung, K.* and Schmerling, E. and Pavone, M.},journal={{IEEE Robotics and Automation Letters}},number={2},pages={295--302},title={{Multimodal Deep Generative Models for Trajectory Prediction: A Conditional Variational Autoencoder Approach}},volume={6},arxiv={2008.03880},category={interaction},img={IvanovicLeungEtAl2020.jpg},selected={true},year={2021}}
On Infusing Reachability-Based Safety Assurance within Planning Frameworks for Human-Robot Vehicle Interactions
Leung, K.,
Schmerling, E.,
Zhang, M.,
Chen, M.,
Talbot, J.,
Gerdes, J. C.,
and Pavone, M.
Action anticipation, intent prediction, and proactive behavior are all desirable characteristics for autonomous driving policies in interactive scenarios. Paramount, however, is ensuring safety on the road—a key challenge in doing so is accounting for uncertainty in human driver actions without unduly impacting planner performance. This paper introduces a minimally-interventional safety controller operating within an autonomous vehicle control stack with the role of ensuring collision-free interaction with an externally controlled (e.g., human-driven) counterpart while respecting static obstacles such as a road boundary wall. We leverage reachability analysis to construct a real-time (100Hz) controller that serves the dual role of (1) tracking an input trajectory from a higher-level planning algorithm using model predictive control, and (2) assuring safety through maintaining the availability of a collision-free escape maneuver as a persistent constraint regardless of whatever future actions the other car takes. A full-scale steer-by-wire platform is used to conduct traffic weaving experiments wherein two cars, initially side-by-side, must swap lanes in a limited amount of time and distance, emulating cars merging onto/off of a highway. We demonstrate that, with our control stack, the autonomous vehicle is able to avoid collision even when the other car defies the planner’s expectations and takes dangerous actions, either carelessly or with the intent to collide, and otherwise deviates minimally from the planned trajectory to the extent required to maintain safety.
@article{LeungSchmerlingEtAl2020,author={Leung, K. and Schmerling, E. and Zhang, M. and Chen, M. and Talbot, J. and Gerdes, J. C. and Pavone, M.},journal={{Int.\ Journal of Robotics Research}},pages={1326--1345},title={{On Infusing Reachability-Based Safety Assurance within Planning Frameworks for Human-Robot Vehicle Interactions}},volume={39},arxiv={2012.03390},category={interaction},img={LeungSchmerlingEtAl2020.png},selected={true},year={2020},issue={10--11}}
Legible and Proactive Robot Planning for Prosocial Human-Robot Interactions
Geldenbott, J.,
and Leung, K.
In Proc. IEEE Conf. on Robotics and Automation,
2024
Humans have a remarkable ability to fluently engage in joint collision avoidance in crowded navigation tasks despite the complexities and uncertainties inherent in human behavior. Underlying these interactions is a mutual understanding that (i) individuals are \textitprosocial, that is, there is equitable responsibility in avoiding collisions, and (ii) individuals should behave \textitlegibly, that is, move in a way that clearly conveys their intent to reduce ambiguity in how they intend to avoid others. Toward building robots that can safely and seamlessly interact with humans, we propose a general robot trajectory planning framework for synthesizing legible and proactive behaviors and demonstrate that our robot planner naturally leads to prosocial interactions. Specifically, we introduce the notion of a \textitmarkup factor to incentivize legible and proactive behaviors and an \textitinconvenience budget constraint to ensure equitable collision avoidance responsibility. We evaluate our approach against well-established multi-agent planning algorithms and show that using our approach produces safe, fluent, and prosocial interactions. We demonstrate the real-time feasibility of our approach with human-in-the-loop simulations.
@inproceedings{GeldenbottLeung2024,author={Geldenbott, J. and Leung, K.},booktitle={{Proc.\ IEEE Conf.\ on Robotics and Automation}},title={{Legible and Proactive Robot Planning for Prosocial Human-Robot Interactions}},year={2024},arxiv={2404.03734},category={interaction},img={GeldenbottLeung2024.png},selected={true},owner={karenl7}}
CoBL-Diffusion: Diffusion-Based Conditional Robot Planning in Dynamic Environments Using Control Barrier and Lyapunov Functions
Mizuta, K.,
and Leung, K.
In IEEE/RSJ Int. Conf. on Intelligent Robots & Systems,
2024
@inproceedings{MizutaLeung2024,author={Mizuta, K. and Leung, K.},booktitle={{IEEE/RSJ Int.\ Conf.\ on Intelligent Robots \& Systems}},title={{{CoBL-Diffusion}: Diffusion-Based Conditional Robot Planning in Dynamic Environments Using Control Barrier and Lyapunov Functions}},year={2024},arxiv={2406.05309},category={interaction},img={MizutaLeung2024.png},selected={true},owner={karenl7}}
Structured Reasoning for Robot Learning
As the field of robot autonomy continues to harness the power of data and deep learning, it also becomes increasingly difficult to explain a robot’s behavior and interpret its decision-making process, necessary prerequisites for robots operating in safety-critical settings.
This line of research develops techniques that infuse structured reasoning into robot learning to enable more interpretable and robust algorithms.
Relevant works
Backpropagation through signal temporal logic specifications: Infusing logical structure into gradient-based methods
This paper presents a technique, named STLCG, to compute the quantitative semantics of Signal Temporal Logic (STL) formulas using computation graphs. STLCG provides a platform which enables the incorporation of logical specifications into robotics problems that benefit from gradient-based solutions. Specifically, STL is a powerful and expressive formal language that can specify spatial and temporal properties of signals generated by both continuous and hybrid systems. The quantitative semantics of STL provide a robustness metric, i.e., how much a signal satisfies or violates an STL specification. In this work, we devise a systematic methodology for translating STL robustness formulas into computation graphs. With this representation, and by leveraging off-the-shelf automatic differentiation tools, we are able to efficiently backpropagate through STL robustness formulas and hence enable a natural and easy-to-use integration of STL specifications with many gradient-based approaches used in robotics. Through a number of examples stemming from various robotics applications, we demonstrate that STLCG is versatile, computationally efficient, and capable of incorporating human-domain knowledge into the problem formulation.
@article{LeungArechigaEtAl2021,author={Leung, K. and Ar\'{e}chiga, N. and Pavone, M.},journal={{Int.\ Journal of Robotics Research}},title={{Backpropagation through signal temporal logic specifications: Infusing logical structure into gradient-based methods}},year={2022},arxiv={2008.00097},category={structure},img={LeungArechigaEtAl2021.png},selected={true}}
Towards the Unification and Data-Driven Synthesis of Autonomous Vehicle Safety Concepts
Leung, K.*,
Bajcsy, A.*,
Schmerling, E.,
and Pavone, M.
As safety-critical autonomous vehicles (AVs) will soon become pervasive in our society, a number of safety concepts for trusted AV deployment have been recently proposed throughout industry and academia. Yet, agreeing upon an "appropriate" safety concept is still an elusive task. In this paper, we advocate for the use of Hamilton Jacobi (HJ) reachability as a unifying mathematical framework for comparing existing safety concepts, and propose ways to expand its modeling premises in a data-driven fashion. Specifically, we show that (i) existing predominant safety concepts can be embedded in the HJ reachability framework, thereby enabling a common language for comparing and contrasting modeling assumptions, and (ii) HJ reachability can serve as an inductive bias to effectively reason, in a data-driven context, about two critical, yet often overlooked aspects of safety: responsibility and context-dependency.
@article{LeungBajcsyEtAl2022,author={Leung, K.* and Bajcsy, A.* and Schmerling, E. and Pavone, M.},journal={{{Available at }\url{https://arxiv.org/abs/2107.14412}}},title={{Towards the Unification and Data-Driven Synthesis of Autonomous Vehicle Safety Concepts}},year={2022},arxiv={2107.14412},category={structure},img={BajcsyLeungEtAl2021.png},selected={true},keywords={preprint}}
Learning Autonomous Vehicle Safety Concepts from Demonstrations
Leung, K.,
Veer, S.,
Schmerling, E.,
and Pavone, M.
Evaluating the safety of an autonomous vehicle (AV) depends on the behavior of surrounding agents which can be heavily influenced by factors such as environmental context and informally-defined driving etiquette. A key challenge is in determining a minimum set of assumptions on what constitutes reasonable foreseeable behaviors of other road users for the development of AV safety models and techniques. In this paper, we propose a data-driven AV safety design methodology that first learns "reasonable" behavioral assumptions from data, and then synthesizes an AV safety concept using these learned behavioral assumptions. We borrow techniques from control theory, namely high order control barrier functions and Hamilton-Jacobi reachability, to provide inductive bias to aid interpretability, verifiability, and tractability of our approach. In our experiments, we learn an AV safety concept using demonstrations collected from a highway traffic-weaving scenario, compare our learned concept to existing baselines, and showcase its efficacy in evaluating real-world driving logs.
@inproceedings{LeungVeerEtAl2023,author={Leung, K. and Veer, S. and Schmerling, E. and Pavone, M.},booktitle={{American Control Conference}},title={{Learning Autonomous Vehicle Safety Concepts from Demonstrations}},year={2023},arxiv={2210.02761},category={structure},img={LeungVeerEtAl2022.png},selected={true},owner={karenl7}}
Learning Responsibility Allocations for Safe Human-Robot Interaction with Applications to Autonomous Driving
Cosner, R.,
Chen, Y.,
Leung, K.,
and Pavone, M.
In Proc. IEEE Conf. on Robotics and Automation,
2023
Drivers have a responsibility to exercise reasonable care to avoid collision with other road users. This assumed responsibility allows interacting agents to maintain safety without explicit coordination. Thus to enable safe autonomous vehicle (AV) interactions, AVs must understand what their responsibilities are to maintain safety and how they affect the safety of nearby agents. In this work we seek to understand how responsibility is shared in multi-agent settings where an autonomous agent is interacting with human counterparts. We introduce Responsibility-Aware Control Barrier Functions (RA-CBFs) and present a method to learn responsibility allocations from data. By combining safety-critical control and learning-based techniques, RA-CBFs allow us to account for scene-dependent responsibility allocations and synthesize safe and efficient driving behaviors without making worst-case assumptions that typically result in overly-conservative behaviors. We test our framework using real-world driving data and demonstrate its efficacy as a tool for both safe control and forensic analysis of unsafe driving.
@inproceedings{CosnerChenEtAl2023,author={Cosner, R. and Chen, Y. and Leung, K. and Pavone, M.},booktitle={{Proc.\ IEEE Conf.\ on Robotics and Automation}},title={{Learning Responsibility Allocations for Safe Human-Robot Interaction with Applications to Autonomous Driving}},year={2023},arxiv={2303.03504},category={structure},img={CosnerChenEtAl2023.png},selected={true},owner={karenl7}}
Learning responsibility allocations for multi-agent interactions: A differentiable optimization approach with control barrier functions
From autonomous driving to package delivery, ensuring safe yet efficient multi-agent interaction is challenging as the interaction dynamics are influenced by hard-to-model factors such as social norms and contextual cues. Understanding these influences can aid in the design and evaluation of socially-aware autonomous agents whose behaviors are aligned with human values. In this work, we seek to codify factors governing safe multi-agent interactions via the lens of responsibility, i.e., an agent’s willingness to deviate from their desired control to accommodate safe interaction with others. Specifically, we propose a data-driven modeling approach based on control barrier functions and differentiable optimization that efficiently learns agents’ responsibility allocation from data. We demonstrate on synthetic and real-world datasets that we can obtain an interpretable and quantitative understanding of how much agents adjust their behavior to ensure the safety of others given their current environment.
@inproceedings{RemyFridovichKeilEtAl2025,author={Remy, I. and Fridovich-Keil, D. and Leung, K.},booktitle={{American Control Conference}},title={{Learning responsibility allocations for multi-agent interactions: A differentiable optimization approach with control barrier functions}},year={2025},arxiv={2410.07409},category={structure},img={RemyFridovichKeilEtAl2024.png},selected={true},keywords={in print},owner={karenl7}}
STLCG++: A Masking Approach for Differentiable Signal Temporal Logic Specification
Signal Temporal Logic (STL) offers a concise yet expressive framework for specifying and reasoning about spatio-temporal behaviors of robotic systems. Attractively, STL admits the notion of robustness, the degree to which an input signal satisfies or violates an STL specification, thus providing a nuanced evaluation of system performance. Notably, the differentiability of STL robustness enables direct integration to robotics workflows that rely on gradient-based optimization, such as trajectory optimization and deep learning. However, existing approaches to evaluating and differentiating STL robustness rely on recurrent computations, which become inefficient with longer sequences, limiting their use in time-sensitive applications. In this paper, we present STLCG++, a masking-based approach that parallelizes STL robustness evaluation and backpropagation across timesteps, achieving more than 1000x faster computation time than the recurrent approach. We also introduce a smoothing technique for differentiability through time interval bounds, expanding STL’s applicability in gradient-based optimization tasks over spatial and temporal variables. Finally, we demonstrate STLCG++’s benefits through three robotics use cases and provide open-source Python libraries in JAX and PyTorch for seamless integration into modern robotics workflows.
@inproceedings{KapoorMizutaEtAl2025,author={Kapoor, P. and Mizuta, K. and Kang, E. and Leung, K.},title={{{STLCG++}: A Masking Approach for Differentiable Signal Temporal Logic Specification}},year={2025},arxiv={2501.04194},category={structure},img={KapoorMizutaEtAl2025.png},selected={true},note={{preprint}},html={https://uw-ctrl.github.io/stlcg/},keywords={submitted},owner={karenl7}}
Real-time decision-making and control for safety-critical operations
Autonomous decision-making in safety-critical scenarios require fast and efficient techniques to understand the environment, reasoning about safety risks, and calculating safe and feasible trajectories. This line of research focuses on developing real-time algorithms and techniques that enable autonomous systems to safely and swiftly operate in safety-critical scenarios, such as autonomous driving and planetary landing.
Relevant works
HALO: Hazard-Aware Landing Optimization for Autonomous Systems
Hayner, C. R.,
Buckner, S. C.,
Broyles, D.,
Madewell, E.,
Leung, K.,
and Açıkmeşe, B.
In Proc. IEEE Conf. on Robotics and Automation,
2023
With autonomous aerial vehicles enacting safety-critical missions, such as the Mars Science Laboratory Curiosity rover’s entry, descent, and landing on Mars, awareness and reasoning regarding potentially hazardous landing sites is paramount. This paper presents a coupled perception-planning solution which address the real-time hazard detection, optimal landing trajectory generation, and contingency planning challenges encountered when landing in uncertain environments. The perception and planning components are addressed by the proposed Hazard-Aware Landing Site Selection (HALSS) framework and Adaptive Deferred-Decision Trajectory Optimization (Adaptive-DDTO) algorithm respectively. The HALSS framework processes point clouds through a segmentation network to predict a binary safety map which is analyzed using the medial axis transform to efficiently identify circular, safe landing zones. The Adaptive-DDTO algorithm address the need for contingency planning during target failure scenarios through adaptively recomputed multi-target trajectory optimization. Overall, Adaptive-DDTO achieves 18.16% increase in terms of landing success rate and 0.4% decrease in cumulative control effort compared to its predecessor, DDTO, while computing near real-time solutions, when coupled with HALSS, in a simulated environment.
@inproceedings{HaynerBucknerEtAl2023,author={Hayner, C. R. and Buckner, S. C. and Broyles, D. and Madewell, E. and Leung, K. and A\c{c}{\i}kme\c{s}e, B.},booktitle={{Proc.\ IEEE Conf.\ on Robotics and Automation}},title={{{HALO}: Hazard-Aware Landing Optimization for Autonomous Systems}},year={2023},arxiv={2304.01583},category={safetycritical},img={HaynerBucknerEtAl2023.png},selected={true},owner={karenl7}}
Receding Horizon Planning with Rule Hierarchies for Autonomous Vehicles
Veer, S.,
Leung, K.,
Cosner, R.,
Chen, Y.,
and Pavone, M.
In Proc. IEEE Conf. on Robotics and Automation,
2023
Autonomous vehicles must often contend with conflicting planning requirements, e.g., safety and comfort could be at odds with each other if avoiding a collision calls for slamming the brakes. To resolve such conflicts, assigning importance ranking to rules (i.e., imposing a rule hierarchy) has been proposed, which, in turn, induces rankings on trajectories based on the importance of the rules they satisfy. On one hand, imposing rule hierarchies can enhance interpretability, but introduce combinatorial complexity to planning; while on the other hand, differentiable reward structures can be leveraged by modern gradient-based optimization tools, but are less interpretable and unintuitive to tune. In this paper, we present an approach to equivalently express rule hierarchies as differentiable reward structures amenable to modern gradient-based optimizers, thereby, achieving the best of both worlds. We achieve this by formulating rank-preserving reward functions that are monotonic in the rank of the trajectories induced by the rule hierarchy; i.e., higher ranked trajectories receive higher reward. Equipped with a rule hierarchy and its corresponding rank-preserving reward function, we develop a two-stage planner that can efficiently resolve conflicting planning requirements. We demonstrate that our approach can generate motion plans in 7-10 Hz for various challenging road navigation and intersection negotiation scenarios.
@inproceedings{VeerLeungEtAl2023,author={Veer, S. and Leung, K. and Cosner, R. and Chen, Y. and Pavone, M.},booktitle={{Proc.\ IEEE Conf.\ on Robotics and Automation}},title={{Receding Horizon Planning with Rule Hierarchies for Autonomous Vehicles}},year={2023},arxiv={2212.03323},category={safetycritical},img={VeerLeungEtAl2023.png},selected={true},owner={karenl7}}
Refining Obstacle Perception Safety Zones via Maneuver-Based Decomposition
Topan, S.,
Chen, Y.,
Schmerling, E.,
Leung, K.,
Nilsson, J.,
Cox, M.,
and Pavone, M.
A critical task for developing safe autonomous driving stacks is to determine whether an obstacle is safety-critical, i.e., poses an imminent threat to the autonomous vehicle. Our previous work showed that Hamilton Jacobi reachability theory can be applied to compute interaction-dynamics-aware perception safety zones that better inform an ego vehicle’s perception module which obstacles are considered safety-critical. For completeness, these zones are typically larger than absolutely necessary, forcing the perception module to pay attention to a larger collection of objects for the sake of conservatism. As an improvement, we propose a maneuver-based decomposition of our safety zones that leverages information about the ego maneuver to reduce the zone volume. In particular, we propose a "temporal convolution" operation that produces safety zones for specific ego maneuvers, thus limiting the ego’s behavior to reduce the size of the safety zones. We show with numerical experiments that maneuver-based zones are significantly smaller (up to 76% size reduction) than the baseline while maintaining completeness.
@inproceedings{TopanChenEtAl2023,author={Topan, S. and Chen, Y. and Schmerling, E. and Leung, K. and Nilsson, J. and Cox, M. and Pavone, M.},booktitle={{IEEE Intelligent Vehicles Symposium}},title={{Refining Obstacle Perception Safety Zones via Maneuver-Based Decomposition}},year={2023},arxiv={2308.06337},category={safetycritical},img={TopanChenEtAl2023.png},selected={true},owner={karenl7}}
Continuous-Time Line-of-Sight Constrained Trajectory Planning for 6-Degree of Freedom Systems
Hayner, C. R.,
Carson, J.,
Açıkmeşe, B.,
and Leung, K.
IEEE Robotics and Automation Letters,
2025
(in print)
Perception algorithms are ubiquitous in modern autonomy stacks, providing necessary environmental information to operate in the real world. Many of these algorithms depend on the visibility of keypoints, which must remain within the robot’s line-of-sight (LoS), for reliable operation. This paper tackles the challenge of maintaining LoS on such keypoints during robot movement. We propose a novel method that addresses these issues by ensuring applicability to various sensor footprints, adaptability to arbitrary nonlinear dynamics, and constant enforcement of LoS throughout the robot’s path. Through our experiments, we show that the proposed approach achieves significantly reduced LoS violation and runtime compared to existing state-of-the-art methods in several representative and challenging scenarios.
@article{HaynerCarsonEtAl2024,author={Hayner, C.~R. and Carson, J. and A\c{c}{\i}kme\c{s}e, B. and Leung, K.},journal={{IEEE Robotics and Automation Letters}},title={{Continuous-Time Line-of-Sight Constrained Trajectory Planning for 6-Degree of Freedom Systems}},year={2025},arxiv={2410.22596},category={safetycritical},img={HaynerCarsonEtAl2024.png},selected={true},note={{(In print)}},html={https://haynec.github.io/papers/los/},keywords={in print},owner={karenl7}}