Safety Assurances for Learning-Enabled Robotic Systems

Speaker

Somil Bansal
Assistant Professor, Electrical and Computer Engineering, University of Southern California

Title

"Safety Assurances for Learning-Enabled Robotic Systems"

Abstract

The ability of machine learning techniques to leverage data and process rich sensory inputs (e.g., vision) makes them highly appealing for use in robotic systems. However, the inclusion of learning-based components in the control loop poses an important challenge: how can we guarantee the safety of such systems?

Control theory provides a number of powerful methods for designing safe controllers for robotic systems, such as Hamilton-Jacobi (HJ) reachability analysis. However, these methods often lack the scalability and flexibility to interface with real-world data and machine-learning models. We present neural reachable tubes that leverage traditional HJ safety conditions within a machine-learning framework to design safe controllers from data. Neural reachable tubes are easily scalable to high-dimensional systems, allowing us to learn safe controllers for a broad range of robotic systems. We will next present a toolbox of methods that can leverage neural tubes to update the safety guarantees online within a fraction of milliseconds as new environment information is obtained during deployment. In the second part of the talk, we will discuss how we can use neural tubes to stress-test learning and vision-based controllers to discover their safety-critical failures and use them to improve the controller. 

Together, these advances provide a continual safety framework for learning-enabled robotic systems, where safety is integrated in different stages of the learning process, starting from their design to their deployment to iteratively improving the system's safety over its lifecycle. Throughout the talk, we will illustrate these methods on various robotic platforms, including autonomous driving, legged locomotion, and navigating in a priori unknown environments.

About Speaker

Somil Bansal is an Assistant Professor in the Department of Electrical and Computer Engineering and the Department of Computer Science at the University of Southern California, Los Angeles. He received a Ph.D. in Electrical Engineering and Computer Sciences (EECS) from the University of California at Berkeley in 2020. Before that, he obtained a B.Tech. in Electrical Engineering from IIT Kanpur, and an M.S. in EECS from UC Berkeley in 2012 and 2014, respectively. After his PhD, he spent a year as a Research Scientist at Waymo (formerly known as the Google Self-Driving Car project). He has also collaborated closely with companies like Skydio, Google, Boeing, NVIDIA, as well as NASA JPL. Somil is broadly interested in developing mathematical tools and algorithms for the control and analysis of safety-critical robotic systems. Somil has received several awards, most notably the NSF CAREER award, the Eli Jury Award for his doctoral research, the RSS Pioneer Award, and the Outstanding Graduate Student Instructor Award.