Jia Li: Collective rationality from self-interested agents in mixed autonomy traffic

Abstract: The rapid development of self-driving technologies makes it imperative to characterize, model, and design the behaviors of mixed autonomy traffic. The problems are especially important, yet more challenging, when automated vehicles are self-interested, instead of being centrally controlled or coordinated. In this talk, I will present my research towards filling the gaps in this direction. I will first introduce an equilibrium model of mixed autonomy traffic based on bargaining games. An intriguing finding is that, under mild conditions, agents in mixed autonomy traffic can always reach Pareto-efficient Nash equilibria, i.e., “collective rationality”, even if they are self-interested and behave heterogeneously. Then I will present a computational model of mixed traffic flow incorporating agent interactions and show how this model achieves an improved simulation accuracy on real-world data. I will conclude the talk with a brief discussion of my research plan.

Biography: Jia Li is an Assistant Adjunct Professor in the Department of Civil and Environmental Engineering, University of California, Davis. He holds a Ph.D. in Transportation from UC Davis. His primary research interest is the modeling, control, and behavior design of automated vehicles in complex mixed autonomy environments leveraging game theory and artificial intelligence tools. He is also interested in applying data science methods to solve emerging problems from built environment. He published more than twenty peer-reviewed papers, including those on the prestigious Transportation Research, ISTTT, and IEEE series. He has led several research projects and secured funding as PI and co-PI from different agencies/programs, including CTECH and NCST (USDOT-sponsored UTC centers), Caltrans, California Senate Bill-1, and HuRRI institute.