A Disturbance Activation Approach to Collision Avoidance Autonomous Driving

A very challenging control task for operating autonomous vehicles is to avoid collision with approaching pedestrians or vehicles while tracking a reference trajectory. This paper proposes a Linear-Quadratic Regulator (LQR)-based collision avoidance trajectory tracking controller for an autonomous driving scenario. In this approach, simply the control module activates a designed disturbance when there is a collision situation. A proper design of the disturbance causes the vehicle to prevent the collision and safely maneuver on the road. The asymptotic stability of this feedback LQR-based controller under an infinite horizon is shown. Simulation experiments are carried out to illustrate the performance of the proposed controller.