Event-triggered MPC for Collision Avoidance of Autonomous Vehicles Considering Trajectory Tracking Performance

Collision avoidance strategy is essential for driving safety of autonomous vehicles. Due to the discrepancy between vehicle motion planning and control, it is challenging to plan and track the collision-free trajectories that are adaptive to trajectory tracking errors and feasible to be executed. The paper puts forward a novel collision avoidance method for autonomous vehicles using the event-triggered MPC algorithm. By developing the triggering conditions as the trajectory tracking errors, vehicle motion planning and control are applied at different frequencies, which reduces the computational cost and guarantees the collision avoidance performances. Furthermore, the collision avoidance constraints, together with vehicle kinematic and dynamic models, are included in the proposed method, so that self-driving vehicles can track planned trajectories and the obstacles can be avoided. Simulation results prove the suggested way can generate and track a feasible trajectory to avoid the obstacle and greatly reduce the computational load.