Increased use of autonomy in safety-critical applications necessitates safety and reliability guarantees to prevent disastrous consequences. Autonomous systems often operate under various sources of uncertainty that make it difficult to predict possible outcomes and provide safety guarantees. Dynamic operating conditions also require these systems to be adaptable for reliable deployment. In this talk, I will discuss an assured runtime safety monitoring technique that draws ideas from machine learning, reachability analysis, and verification techniques to assess the safety of an autonomous robot under disturbance. Additionally, I will introduce meta-learning-based approaches that enable autonomous robots to adapt to faulty conditions and improve performance. Finally, I will conclude my talk with a discussion of future research directions.
Esen Yel is a Postdoctoral Scholar at Stanford University affiliated with Stanford Intelligent Systems Lab and Stanford Center for AI Safety. She received her Ph.D. in System Engineering from the University of Virginia in 2021. Her research focuses on planning, runtime monitoring, and online adaptation to improve safety in autonomous systems under uncertainty. She was a participant in the EECS Rising Stars Workshop in 2022 and the Robotics Science and Systems Pioneers Workshop in 2021. At the University of Virginia, she received the Link Lab Outstanding Graduate Research award and the Ruthie Oxford Memorial award. Esen obtained her M.S. and B.S. degrees in Electrical Engineering from Bogazici University in Turkey.