Talk 3

Abstract: 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 talk, I will discuss recent work on learning reasonable and responsible driving behaviors from data and using the learned model for safe control synthesis and safety evaluation for AVs. We leverage control barrier functions as inductive bias when modeling the set of controls that human drivers operate with, and show that it provides interpretability into the safety evaluation of driving scenes and computation of associated safe controls.

Bio: Dr. Karen Leung is an Assistant Professor of Aeronautics and Astronautics at the University of Washington, where she is the director of the Control and Trustworthy Robotics Laboratory. Before joining the University of Washington, she was a Research Scientist within the Autonomous Vehicle Research Group at NVIDIA (and currently retains a part-time appointment). She received her Ph.D. degree in Aeronautics and Astronautics from Stanford University in 2021. Her main research interests are in the development of safe, intelligent, and trustworthy robotic systems, especially those that interact with humans and operate in safety-critical settings. She is a recipient of the UW + Amazon Science Hub Faculty Research Award, the William F. Ballhaus Prize for best Ph.D. Thesis, and the Qualcomm Innovation Fellowship (US).