An adaptive path tracking controller for autonomous vehicles based on the Pure Pursuit algorithm

Aiming at the problem of unmanned vehicle path tracking, this paper proposes an adaptive path tracking controller based on the Pure Pursuit(PP) method. The controller replaces the traditional manual selection of the lookahead distance in the traditional PP algorithm with the Deep Q Network (DQN) algorithm. The controller dynamically adjusts the lookahead distance based on lateral error, heading error, and vehicle speed to adapt to different operating conditions and improve path tracking performance. This paper obtained the Deep Q Network - Pure Pursuit (DQN-PP) adaptive controller model through reinforcement learning training. In order to verify the control effect of the DQN-PP adaptive controller, this paper designed a simulation experiment and analyzed the experimental results. The results show that compared with the traditional method, the DQN-PP adaptive controller has better path tracking performance, and the average error value of path tracking has been reduced by 21%. This paper provides an effective adaptive solution for path tracking control in autonomous vehicles and has practical application value.