Learning, Control, and Safety for Enhanced Legged Dexterity

ABSTRACT

With the advance in optimal control and reinforcement learning, latest controllers have demonstrated exceptional performance on complex multi-limbed robots, including the application in aerial manipulation, quadrupedal locomotion, and dexterous hand skills. Despite these advancements, comprehensive safety validation remains a prerequisite for their large-scale real-world deployment. Current state-of-the-art (SOTA) controllers exhibit robustness against standard testing paradigms, domain randomization, and evaluations by human experts.

In this talk, I will first introduce our progress on enhancing dexterity of the robots, and then our recent research revealing vulnerabilities in long-tested state-of-the-art controllers on the legged locomotion and dexterous manipulation when subjected to minor perturbations generated by AI agents. This study highlights critical safety concerns and emphasizes the necessity of addressing these vulnerabilities to enhance system reliability. The failure cases identified in our analysis offer valuable insights into system components, providing a foundation for improving the robustness and safety of black-box neural controllers.

Fan Shi is an Assistant Professor in National University of Singapore with NUS Presidential Young Professorship. He was previously a Postdoc Researcher working with Prof. Marco Hutter and Prof. Stelian Coros in ETH Zurich. His research focused on learning and control for multi-limbed robots in locomotion and manipulation. He obtained his Master and Ph.D. degree from the Univ. of Tokyo supervised by Prof. Masayuki Inaba, and Bachelor degree from Peking University. He has received several highly competitive awards, such as the Dean’s Award for his Doctoral Thesis, IEEE RAS/JJC Young Award in ICRA 2020, ICRA 2018 Best Paper Award on UAV and AI Safety Finalist Award in Switzerland. More details about his work can be found at: https://fanshi14.github.io/me.