Research On Path Planning Design And Tracking Control Method Of Automatic Parking System

At present,the automobile industry is developing towards the direction of vehicle intelligence.The automatic parking system,as one of the representative systems in the era of vehicle intelligence,can improve the safety and convenience of parking.It integrates perception,planning and execution,and it is an intelligent driving system that can realize automatic driving under local conditions.The automatic parking system is a manifestation of the intelligent research and development capabilities of major auto manufacturers.At the same time,with the frequent parking problems in major cities,consumers' increasing demand for automatic parking systems has further promoted the development of automatic parking systems.The automatic parking system is mainly composed of three core parts: parking space detection,path planning and path tracking.In this paper,combined with the parameters of a certain vehicle model,the three core problems of the automatic parking system are studied.In terms of parking space detection,this paper analyzes the shortcomings of parking space detection based on ultrasonic radar,and uses VLP-16 lidar as the detection sensor for parking space detection.First of all,this paper selects a reasonable point cloud interval;the second step uses RANSAC pavement cutting method to extract pavement point cloud information;then project rasterize the point cloud information,use raster height information to filter out the range of roadside boundary,and use RANSAC to fit the roadside point cloud to get the roadside boundary function;then,the obstacle point cloud data within the boundary of the roadside is selected,and the clustering segmentation based on the DBSCAN algorithm is performed;finally,this paper uses the rectangular frame to select different obstacles,and obtains the distance between adjacent obstacles and the maximum distance from the outer boundary of the rectangle to the roadside,so as to obtain the size information of the parking space.Verification shows that the parking space detection parameters can be detected by the method of this paper for parking space detection in parallel parking conditions on the roadside.In terms of parking path planning,the paper first analyzes the vehicle kinematics model of parking motion,and then uses the reverse driving method to plan the initial parking path and the area of the starting point of the parking;then use the fifth degree polynomial and fourth degree polynomial to smooth the preliminary planned parallel parking path and vertical parking path respectively,and obtain a parking path with continuous curvature and meeting the obstacle avoidance requirements.In view of the small vertical parking spaces,the path function still does not meet the steering constraints after polynomial smoothing.In this paper,the DCD intermediate adjustment trajectory is used,and finally a three-stage vertical parking path with continuous curvature is obtained.In terms of path tracking,this paper first designs a path tracking controller based on nontemporal reference frame,using anti-saturation sliding mode variable structure control;then uses nonlinear fal function approach law to improve the approach quality of sliding mode control;then considers the speed disturbance and steering hysteresis,a linear extended state observer is introduced to compensate for the disturbance,and simulation experiments verify that the controller has a good tracking effect on the sinusoidal path;finally,based on the carsim / simulink joint simulation,the effectiveness of the path tracking controller based on the sliding mode control method is verified.