<p>This paper proposes a lane change obstacle avoidance trajectory planning strategy for autonomous vehicles based on D-APF. The establishment of a safe distance model can avoid extreme behavior of self-driving cars. The improved obstacle repulsion field that combines the safe distance model and the speed difference can effectively improve the comfort of the algorithm planning. The establishment of Frenet- Serret improves the practicability of the algorithm in complex scenarios. Numerical optimization using convex space will further improve the comfort and safety of the path while ensuring the real-time performance of the algorithm. Speed planning solves the appropriate speed curve by considering the dynamic constraints of the vehicle. The obstacle avoidance effect of the algorithm in static and dynamic complex scenes is tested by real-time simulation on the HIL platform. The results show that the proposed algorithm can plan a safe and comfortable obstacle avoidance trajectory in complex scenes under structured roads.</p>