Harnessing Mean-Field Game & Data Science for Mixed Autonomy

Xuan Di
Columbia University
Civil Engineering & Engineering Mechanics

As this era’s biggest game-changer, autonomous vehicles (AV) are expected to exhibit new driving and travel behaviors, thanks to their sensing, communication, and computational capabilities. However, a majority of studies simply tailor human-driven vehicles (HV)’s behavior for AVs by tweaking some behavioral parameters. In these models, AVs are essentially human drivers but react faster, “see” farther, and “know” the road environment better. We believe AVs’ most disruptive characteristic lies in its intelligent goal-seeking and adapting behavior. Based on whether the mixed traffic environment is deterministic or stochastic, we propose two types of controls: game-based and learning-based. I will first introduce a game-theoretic decision-making process for a large number of AVs. To illustrate the potential advantage that AVs may bring to stabilize traffic, we propose a multi-class game where AVs are modeled as intelligent game-players and HVs are modeled using a classical non-equilibrium traffic flow model. I will then briefly touch on our on-going work about a learning-based control when the mixed traffic environment contains uncertainty, which allows AVs to interact with the environment and learn optimal driving policies dynamically.

Presentation (PDF File)

Back to Workshop I: Individual Vehicle Autonomy: Perception and Control