An obstacle avoidance trajectory selection for autonomous vehicles based on improved artificial potential field method

In the urban road network environment, the auto drive system faces the problem of trajectory planning under the coupling of multi-source obstacles. Therefore, a method for obstacle avoidance trajectory selection of autonomous driving vehicles based on improved artificial potential field method is proposed. Build dynamic windows based on kinematic models, discretize velocities to generate feasible velocity sets. Improving the artificial potential field method: introducing a spacing adjustment factor to solve the problem of unreachable targets, designing an elliptical repulsion field to reduce adjacent lane interference, integrating relative velocity repulsion to enhance dynamic obstacle avoidance, and constructing a lane center potential field to suppress path deviation. Finally, generate a smooth path using fourth-order cubic B-spline curves and select the optimal trajectory. The experimental results show that the obstacle avoidance success rate of our method reaches over 95%, and the path selection delay is controlled within the range of 85-97ms.