Mitigating Congestion with Auctions: A Game Theoretic Approach for Selfish Autonomous Vehicles

This study focuses on the high information recognition capability of autonomous vehicles and analyzes the effect of information asymmetry between autonomous and manually operated vehicles on traffic conditions when there is a mixed flow of autonomous and manually operated vehicles. By constructing a lane change model that describes the overtaking behavior as an extensive-form game, it is possible to express the anticipation of the future state and the asymmetry of information. Furthermore, using the lane change model described above, we propose a mechanism to improve the traffic condition by providing appropriate incentives to autonomous vehicles through an auction mechanism. The proposed mechanism has efficiency, Bayesian-Nash-Incentive-Compatibility, and interim individual rationality, and is applicable to mixed situations where some vehicles do not participate in the auction.