Kehan Long

I am currently a PhD candidate in Mathematics and Contextual Robotics Institute at University of California, San-Diego. I work at the Existential Robotics Laboratory. I am fortunate to be advised by Prof. Nikolay Atanasov and Prof. Melvin Leok. I also work closely with Prof. Jorge Cortรฉs.

My academic journey began at the University of Illinois,Urbana-Champaign, where I obtained a B.S. degree in Applied Mathematics, complemented by a minor in Computer Engineering.

Email  /  Google Scholar  /  Github  /  LinkedIn

profile photo

Research Interests

My research interests lie at the intersection of robotics, machine learning, and control theory, with an emphasis on safe autonomy. My work focuses on developing safety and stability verification methodologies for robotic control systems amid model and environmental uncertainties. My goal is to devise scalable and robust approaches for operating autonomous systems within highly dynamic and uncertain settings, while providing safety and stability guarantees.

News

[May 2024] Our paper on Sensor-based Distributionally Robust Control for Safe Robot Navigation in Unknown Dynamic Environments has been submitted to IJRR.

[April 2024] Our paper on Distributionally Robust Policy and Lyapunov-Certificate Learning has been submitted to IEEE OJ-CSYS.

[Dec 2023] Our paper on Feasibility Analysis and Regularity Characterization of Distributionally Robust Safe Stabilizing Controllers has been accepted to L-CSS. ๐ŸŽ‰๐ŸŽ‰๐ŸŽ‰

[March 2023] Our paper on Distributionally Robust Lyapunov Function Search Under Uncertainty has been accepted to L4DC 2023. ๐ŸŽ‰๐ŸŽ‰๐ŸŽ‰

[Jan 2023] Our paper on Safe and Stable Control Synthesis for Uncertain System Models via Distributionally Robust Optimization has been accepted to ACC'23. ๐ŸŽ‰๐ŸŽ‰๐ŸŽ‰

[Nov 2022] Our paper on Distributionally Robust Lyapunov Function Search Under Uncertainty has been submitted to L4DC 2023.

[Sep 2022] Our paper on Safe and Stable Control Synthesis for Uncertain System Models via Distributionally Robust Optimization has been submitted to ACC'23.

[June 2022] Our paper on Safe Control Synthesis with Uncertain Dynamics and Constraints has been accepted to RA-L with IROS 2022. ๐ŸŽ‰๐ŸŽ‰๐ŸŽ‰

[March 2021] Our paper on Learning Barrier Functions with Memory for Robust Safe Navigation has been accepted to RA-L with ICRA 2021. ๐ŸŽ‰๐ŸŽ‰๐ŸŽ‰

Selected Publications

Sensor-based Distributionally Robust Control for Safe Robot Navigation in Unknown Dynamic Environments


Kehan Long, Yinzhuang Yi, Zhirui Dai, Sylvia Herbert, Jorge Cortรฉs, Nikolay Atanasov
submitted to The International Journal of Robotics Research (IJRR), 2024
arxiv / code / website

We introduce a novel method for safe and efficient navigation of mobile robots in dynamic, unknown environments, utilizing onboard sensing without the need for accurate map reconstructions. Traditional methods typically rely on detailed map representations to synthesize safe stabilizing controls for mobile robots, which can be computationally demanding and less effective, particularly in dynamic scenarios. By leveraging recent advances in distributionally robust optimization, we propose a novel distributionally robust control barrier constraint (DR-CBC) formulation that directly processes range sensor data to ensure safe autonomy. Together with a control Lyapunov function for path tracking, we demonstrate the superiority of our formulation in robot navigation, showcasing superior performance in terms of safety, efficiency, and robustness against sensory and environmental uncertainties. Both simulated and real-world experiments validate the effectiveness of our approach, marking a substantial advancement in real-time safe autonomy for mobile robots in dynamic and unknown environments.

project image
project image

Distributionally Robust Policy and Lyapunov-Certificate Learning


Kehan Long, Jorge Cortรฉs, Nikolay Atanasov
submitted to IEEE Open Journal of Control Systems (OJ-CSYS), 2024
arxiv / code

This article presents novel methods for synthesizing distributionally robust stabilizing neural controllers and certificates for control systems under model uncertainty. A key challenge in designing controllers with stability guarantees for uncertain systems is the accurate determination of and adaptation to shifts in model parametric uncertainty during online deployment. We tackle this with a novel distributionally robust formulation of the Lyapunov derivative chance constraint ensuring a monotonic decrease of the Lyapunov certificate. To avoid the computational complexity involved in dealing with the space of probability measures, we identify a sufficient condition in the form of deterministic convex constraints that ensures the Lyapunov derivative constraint is satisfied. We integrate this condition into a loss function for training a neural network-based controller and show that, for the resulting closed-loop system, the global asymptotic stability of its equilibrium can be certified with high confidence, even with Out-of-Distribution (OoD) model uncertainties. To demonstrate the efficacy and efficiency of the proposed methodology, we compare it with an uncertainty-agnostic baseline approach and several reinforcement learning approaches in two control problems in simulation.

project image

Distributionally Robust Lyapunov Function Search Under Uncertainty


Kehan Long, Yinzhuang Yi, Jorge Cortรฉs, Nikolay Atanasov
5th Learning for Dynamics & Control Conference (L4DC), 2023
arxiv / code

This paper devises methods for proving Lyapunov stability of dynamical systems subject to disturbances with an unknown distribution. We assume only a finite set of disturbance samples is available and that the true online disturbance realization may be drawn from a different distribution than the given samples. We formulate an optimization problem to search for a sum-of-squares (SOS) Lyapunov function and introduce a distributionally robust version of the Lyapunov function derivative constraint. We show that this constraint may be reformulated as several SOS constraints for polynomial systems. For general and higher dimensional systems, we provide a distributionally robust chance-constrained formulation for neural network Lyapunov function search.

project image

Safe Control Synthesis With Uncertain Dynamics and Constraints


Kehan Long, Vikas Dhiman, Melvin Leok, Jorge Cortรฉs, Nikolay Atanasov
IEEE Robotics and Automation Letters (RA-L), 2022
arxiv

This paper explores the synthesis of safe controls for dynamical systems with either probabilistic or worst-case uncertainty in both the dynamics model and the safety constraints (environments). We formulate novel probabilistic and robust (worst-case) control Lyapunov function (CLF) and control barrier function (CBF) constraints that take into account the effect of uncertainty in either case. We show that either the probabilistic or the robust (worst-case) formulation leads to a second-order cone program (SOCP), which enables efficient safe stabilizing control synthesis in real-time.

project image

Learning Barrier Functions with Memory for Robust Safe Navigation


Kehan Long*, Cheng Qian*, Jorge Cortรฉs, Nikolay Atanasov
IEEE Robotics and Automation Letters (RA-L), 2021
arxiv

This paper investigates safe navigation in unknown environments, using on-board range sensing to construct control barrier functions online. To represent different objects in the environment, we use the distance measurements to train neural network approximations of the signed distance functions incrementally with replay memory. This allows us to formulate a novel robust control barrier safety constraint which takes into account the error in the estimated distance fields and its gradient. Our formulation leads to a second-order cone program, enabling safe and stable control synthesis in a prior unknown environments.




Contact

You are very welcome to contact me regarding my research and collaboration. I can be contacted directly at kehan.lkh@gmail.com



The template is borrowed from Jon Barron's website and Leonid Keselman's Jekyll fork.