Cooperation of autonomous vehicles using a hierarchy of auction-based and model-predictive control

This paper proposes an approach to establish cooperative behavior within traffic scenarios involving only autonomously driving vehicles. The main idea is to employ principles of auction-based control to determine driving strategies by which the vehicles reach their driving goals, while adjusting their paths to each other and adhering to imposed constraints like traffic rules. Driving plans (bids) are repetitively negotiated among the control units of the vehicles (the auction) to obtain a compromise between separate (local) vehicle goals and the global objective to resolve the considered traffic scenario. The agreed driving plans serve as reference trajectories for local model-predictive controllers of the vehicles to realize the driving behavior. The approach is illustrated for a cooperative overtaking scenario comprising three vehicles.