Abstract: Path Tracking plays important role in the lateral control of autonomous vehicles. The stability and tracking accuracy are usually related to vehicle speed, road curvature, etc., which directly affect the safety in complex driving conditions. To improve the stability and tracking accuracy under complex conditions, path planning, tracking control and stability control are combined together to design a tracking control method based on adaptive preview paths. First, based on the vehicle's two-degree-of-freedom model, a preview distance adaptive algorithm is designed, which adjusts the preview distance according to the vehicle dynamics state and road adhesion. Secondly, the preview path at the desired preview distance is given by a cubic polynomial fitting method. Finally, based on performance of obstacle avoidance, tracking accuracy, and vehicle stability, a particle swarm optimization algorithm(PSO) is designed to optimize the algorithm parameters. The performances in path tracking, lane changing and obstacle avoidance conditions are verified in the hardware-in-the-loop tests and vehicle tests. The results show that the algorithm can adaptively adjust the preview path during tracking with low computation burden, and achieve the balance of tracking accuracy and vehicle stability.