Path Planning And Controlalorithm Of Home-zone Parking Assist System

With the rapid development of automobile electronization and advanced auxiliary driving technology,automatic driving,as an advanced stage of auxiliary driving technology,has become an important way to solve traffic problems in the future,and has become a new technology research hotspot and focus in the world.The automatic parking system can liberate the driver in the final parking stage of driving,and the self-learning parking system is the advanced stage of automatic parking.It relies on the fusion of vision,radar and other multi-sensor to realize the rapid mapping and positioning in the parking lot and the path learning of the driver in the parking process.Aiming at the path planning and tracking control algorithm of self-learning parking system,the overall planning control scheme is determined,and the theoretical basis and mathematical model of each module are analyzed.This paper presents a complete algorithm design and program implementation of path planning and tracking control module of self-learning parking system,including path planning algorithm,longitudinal control algorithm and transverse control algorithm.Path planning is divided into two parts: Parking tracking planning and parking warehousing planning;The optimal path planning algorithm is designed based on dynamic programming and quadratic programming;And the design of storage algorithm based on hybrid a * algorithm.The traditional longitudinal control algorithm does not consider the response process of the bottom braking system,which can not meet the parking accuracy of automatic parking demand.To solve this problem,a lpc-pid cooperative longitudinal control scheme is proposed,which is based on an advanced longitudinal predictive control algorithm(LPC).LPC algorithm can actively predict the braking distance,and give consideration to speed planning and longitudinal control.The dynamic model of the braking system and the kinematic model of the whole vehicle are established,and the index of the longitudinal braking response is proposed.The braking response model is identified by theoretical derivation and real vehicle test.The current lateral control algorithm for medium curvature and large curvature path tracking error,poor timeliness,is due to the use of self-learning parking system scene characteristics of path curvature,low speed.To solve this problem,a new curvature adaptive model predictive control algorithm is proposed in this paper,which can improve the adaptability and robustness of MPC controller to curvature while ensuring the timeliness.Its adaptability to curvature mainly comes from the new curvature feed-forward compensator and the curvature adaptive structure of cost function weight.Based on ROS as the communication middleware,the program development,compilation and deployment are completed.The simulation of path planning and tracking control nodes is completed in the simulator,and the real vehicle closed-loop test is carried out.After verification,the parking accuracy of advanced longitudinal predictive control algorithm is less than 10 cm;The lateral error of the adaptive model predictive control algorithm is less than 20 cm.Compared with the standard MPC controller,the steady-state lateral error is reduced by 51%,and the peak lateral error is reduced by 46%.