Abstract: Aiming at the problem of steering collision avoidance for autonomous vehicle, a hierarchical control method for collision avoidance is proposed. In upper-layer controller for path planning, the artificial potential field function is introduced based on vehicle kinematic model and the relative state between vehicle and obstacle is used to describe the collision risk of vehicle. While in bottom-layer controller for path tracking, the nonlinear kinetic model of vehicle and the path tracking optimization function based on optimal steering wheel angle input are established to achieve local collision avoidance path tracking. Finally Carsim/Matlab joint simulation platform is constructed to simulate the collision avoidance path planning and tracking effects of controlled vehicle under different speeds and road conditions. The results show that the upper controller can plan local collision avoidance path real time according to obstacle information, while the bottom-layer controller can control the vehicle to smoothly and stably track reference path, hence fulfilling the active collision avoidance function of vehicle.

Key words: autonomous vehicle, hierarchical control, active collision avoidance, path planning, path tracking