Design of DQN-Based Path Tracking Algorithm for Robust Autonomous Driving

Despite the increasing interest in autonomous driving, significant challenges remain, particularly in achieving high and stable tracking performance. We herein propose a deep Q-network (DQN)-based path tracking algorithm to enhance autonomous vehicle navigation. By interpreting the path tracking task as a sequential decision-making problem, we develop a DQN-based steering control algorithm for precise path tracking. Simulation results confirm that the proposed algorithm significantly enhances tracking accuracy and stability, out performing the conventional methods.