Abstract: As the scale of autonomous operations grows in aerospace, an autonomous aircraft or spacecraft's performance will be significantly impacted by resource competition and interaction uncertainty between vehicles. Combining Markov decision process, game theory, and optimization, my research aims to intelligently coordinate multi-vehicle trajectory planning to optimize fleet-level performance. In this talk, I will introduce the Markov decision process congestion game model to resolve congestion in air traffic management and ride-hail. Next, building on robust dynamic programming, I will analyze how uncoordinated vehicle interactions impact the learning dynamics of an individual vehicle and derive Hausdorff distance convergence results for diverging value iteration. I will end by discussing some exciting domains where resource-centric and intelligent multi-vehicle coordination is applicable.
Bio: Sarah H.Q. Li is a Ph.D. candidate in Aeronautics and Astronautics Engineering at the University of Washington and has received her B.A.Sc. in Engineering Physics with a minor in Mathematics from the University of British Columbia. Her research combines game theory, stochastic control, and optimization to enable large-scale autonomous interactions in disruption-prone environments such as roadways and airspaces. She is a 2020 Zonta International Amelia Earhart Fellow and a 2022 UW Aero&Astro Condit Fellow.