Edge-assisted cooperative control and flexible formation for connected and automated vehicles

Vehicle platoons have many potential advantages, such as improving road capacity, enhancing traffic efficiency, and saving energy. It is one of the typical application scenarios of multi-vehicle systems achieving flexible formation in which scattered vehicles join the platoon through active or forced lane change. Considering the limitation of on-board computing and communication resources, and the complexity of cooperative lane changing and formation tasks, this paper uses the edge server to realize the cooperative process of vehicles joining the platoon through lane changing. Specifically, the edge server collects the state information and destination data of all vehicles, calculates the optimal lane-changing trajectory for the lane-changing vehicle, and designs its tracking algorithm based on model predictive control. On the other hand, based on the distributed deep deterministic strategy gradient algorithms, the edge server carries out cooperative adaptive cruise control for the platoon, coordinates the platoon to generate the safety interval, and sends out the cooperative lane-changing suggestions to the lane-changing vehicle and the platoon to realize the safe and efficient joining of the lane-changing vehicle into the platoon. In particular, reasonable reinforcement learning reward functions and safe spacing generation policies are designed to achieve secure, stable, and fast platoon control, and suppress the acceleration and velocity oscillation when the lane-changing vehicle joins the platoon. Finally, compared with various multi-vehicle cooperative strategies, the superiority of the proposed control strategies and cooperative schemes are verified through the joint simulations.