Adaptive path planning for autonomous parking based on hybrid PSO-GA optimization algorithm

Aiming at the parking scene of autonomous parking in vertical parking scene, an adaptive path planning algorithm is proposed. First, a path planning method based on curve interpolation is introduced. According to the selected parking space, the parking termination point is determined, and the reverse parking path is planned sequentially using straight line, clothoid, arc, and fifth-degree polynomial curve. By considering the steering constraints of the vehicle and collision constraints during the parking process, variables such as straight line length, arc length, and arc radius are optimized within specified boundary intervals. Cost functions and a fitness function are established based on the smoothness, economy, and space utilization of the path, and the generated path is optimized using a hybrid PSO-GA combined with the path planning algorithm. Finally, MATLAB is used to simulate the parking paths generated by the proposed algorithm and the hybrid A* algorithm. The results indicate that the proposed adaptive path planning algorithm based on the hybrid PSO-GA has a continuous curvature compared to the path planned by the hybrid A* algorithm. Under the same path resolution, the proposed algorithm reduces computation time by 88.33% compared to the hybrid A* algorithm. This method not only resolves the issue of abrupt curvature changes but also reduces computation time, thereby improving parking planning efficiency.