A Parking Pose and Trajectory Selection Algorithm Based on Artificial Potential Field and Particle Swarm Optimization

Introduction:Path planning is one of the fundamental issues for automated valet parking system.However, current automatic parking system solutions simply choose the center of the parking bay as target parking pose, which is not the most suitable pose in many situations.Besides, it requires further study to effectively evaluate the parking trajectory; particularly, the path generator might not figure out the suitable trajectory if adjacent vehicle is improperly parked.This paper presents an optimal parking pose and trajectory selection approach based on the information of the ultrasonic radar and vision.To evaluate the positions of obstacles and parking lines when parking, we construct a virtual potential field that can effectively represent the real parking scenario.Thereafter, particle swarm optimization approach is used to iterate possible parking poses and select the one with the least risk.The parking trajectory is generated and optimized to minimize the risk degree, and finally smoothed using spline interpolation.The experiment results show that our method can adaptively adjust the parking pose in different parking scenarios, and able to generate feasible and smooth parking trajectory from arbitrary starting pose to the optimal parking pose.