Trajectory Planner and Motion Control Solution for Automated Valet Parking Function

Abstract

AVL Company has been developing for several years autonomous driving technologies. This article describes the scope of trajectory planning and motion control algorithm and respective performances for the Automated Valet Parking (AVP) function. This function has been developed on an AVL vehicle demonstrator in Shanghai Technical Center (STC). The trajectory planner allows planning in structured environment given reference waypoints. During shuttling, typical Open Planner algorithm is applied for trajectory generation and selection. During parking maneuver, a combination of Hybrid A Star and Reeds-Shepp algorithm are used. Real time improvement of such combination is measured and compared with pure Hybrid A Star approach. Evaluation is performed both on Model in Loop (MiL) and on real vehicle. Additionally, a method to assess the real tests performance of the AVP function on AVL demonstrator vehicle using two different motion controllers is herein presented in order to provide a complete synthesis of challenges related to motion control.