In view of such flaws as low accuracy and poor robustness of target path tracking for autonomous vehicles in motion, a reward function-deep deterministic policy gradient (RF-DDPG) path tracking algorithm has thus been proposed. Based on the deep reinforcement learning DDPG, the algorithm designs the reward function of the DDPG algorithm for an optimization of the DDPG parameters so as to achieve the required tracking accuracy and stability. A simulation experiment has been conducted on the original DDPG algorithm and the improved RF-DDPG path tracking control algorithm based on the aopllo autonomous driving simulation platform. The results show that the proposed RF-DDPG algorithm is characterized with an adavantage over the DDPG algorithm in path tracking accuracy and robust performance.