Research On Optimal Parking Space Allocation And Path Planning For Automated Valet Parking System Considering Mixed Traffic

With the development of autonomous vehicle technology,‘Automated Valet Parking System(AVP)’ is introduced as the representative application of low-speed autonomous driving technology in constraint environment.AVP is developed based on Automated Parking Assistant System(APA)and can help drivers on difficult parking.AVP can automatically drive the vehicle into the parking infrastructure,find the available parking space and reach the target parking space,saving time for drivers.With the continuous iteration of autonomous driving technology,parking space allocation and reasonable path planning under mixed traffic scenario(which is,there are both autonomous vehicles and human-operated vehicles in a parking infrastructure)are the technical difficulties of AVP.In this paper,the optimal parking space allocation and path planning in a parking infrastructure under mixed traffic scenario based on AVP are studied.The main contents are as follows:(1)All the available parking spaces in a parking infrastructure are assigned with weights by Fuzzy Comprehensive Evaluation method.In order to avoid the contradiction between the parking space allocated to the autonomous vehicle and the one selected by the human-operated vehicle,the weights of available parking spaces are assigned and sorted through the analysis of drivers’ parking preference.Through the ranking of weights,the choice of parking space for the human-operated vehicle is predicted and another available parking spaces is allocated to the autonomous vehicle.(2)The global path from the entrance of the parking infrastructure to the allocated parking space is planned through Floyd algorithm.The description of distance is simplified by numbering the nodes of all intersections and parking spaces.According to the traffic rules in the parking infrastructure(road direction),the adjacency matrix between nodes is established to calculate and obtain the most efficient global path.(3)Based on Frenet frame,the local obstacle-avoiding motion planning of vehicle is decoupled into lateral and longitudinal.The position of vehicle can be better described under Frenet frame compared to Cartesian coordinate system.Combined with the position of obstacles and the width of the lane in the parking infrastructure,the lateral and longitudinal optimal trajectory is planned respectively.Trajectories in both directions are coupled then.The optimal local obstacle-avoiding trajectory is obtained by satisfying factors such as the kinematic of the vehicle and the limited curvature of the trajectory.The simulation result shows that the proposed method has good generalization and robustness.