Hierarchical Approach for Safety of Multiple Cooperating Vehicles

In this paper we present a hierarchical structure consisting of offline and online verification, ensuring the safety properties of cooperating vehicles; communicating either implicitly (e.g.: vehicle on left lane slowly opens gap, indicating the vehicle on right lane to merge in front of it) or explicitly via Car-to-Car communication with each other. The offline verification is based on the concept of reachability analysis of hybrid systems and aims at building formally correct and safe cooperative maneuvers for a group of vehicles. The online verification layer has the task of negotiating the possible cooperative maneuvers with other traffic participants and send the selected optimal plan to a low-level tube based Model Predictive Control (MPC). The MPC then calculates the control inputs to be applied to the actuators in order to guide the vehicle safely under the presence of model uncertainty and disturbance. In case MPC offers no feasible solution to the constraint optimization problem i.e. there occurs a constraint violation during the prediction horizon, the safety of the cooperative maneuver is ensured by an emergency planner, which aborts the current cooperative maneuver and brings the vehicle to safe state. The effectiveness and the performance of the hierarchical concept presented here are shown with an exemplary cooperative lane change scenario involving multiple vehicles.