Research On Automatic Parking Path Planning And Tracking Based On Hierarchical Control

The rapid growth of the number of cars has caused parking spaces in cities to become very narrow,resulting in drivers often colliding when parking,and the problem of parking difficulties for vehicles is becoming more and more serious in people's lives.Automatic parking is an important function of intelligent assisted driving,which can effectively reduce the occurrence of parking accidents and is a solution to increase driver comfort and safety.The main research work of this article is as follows:(1)The research proposes a fusion method of parking space marking line detection based on panoramic vision(AVM)and space parking space detection based on lidar.AVM-based parking space marking line detection method obtains panoramic images around the vehicle through the AVM system,and uses the random sampling consistency(RANSAC)method and the chamfer matching(DT)method to estimate the parking space wires and dividers in the image,and uses the parking space marking in the image Line information to detect parking spaces.The fusion method of space parking space detection based on Lidar uses the point cloud information scanned by lidar to detect free space parking spaces by identifying the location information of adjacent vehicles.The fusion method can make up for the limitation of the single sensor to detect the parking space,solve the problem that the parking space cannot be detected without the presence of adjacent vehicles or the parking space marking line is seriously damaged,and improve the detection accuracy of the parking space.(2)The research proposes a parking path planning method based on B-spline theory.A B-spline curve is used to plan the vertical parking path,and the single-step vertical parking and two-step vertical parking methods are adopted according to the space constraints of the parking channel,and the minimum of the single-step vertical parking and the two-step vertical parking method are analyzed at the same time.Parking restrictions.The initial position of the parking path allows the vehicle to start in any initial posture,which effectively reduces the vehicle's steering adjustment at the initial point,and allows the vehicle to recalculate the new path without adjusting the vehicle when it deviates from the planned path.The experimental results show that the curvature of the planned parking path is smooth and easy to track,which meets the minimum turning radius constraints of the vehicle steering actuator and the obstacle avoidance requirements of the parking path.(3)The research proposes a parking path tracking method based on the model predictive control(MPC)algorithm,and designs a parking path tracking controller.Firstly,the vehicle kinematics model is established,then the target optimization function and model constraint function of the path tracking controller are constructed,and the parking path tracking controller is designed according to the established model function.And based on Prescan and Simulink software,a parking path tracking control joint simulation platform was built.The results show that the parking path tracking system designed in this paper can achieve automatic parking better.And compared with the pure tracking path controller.The experimental results show that MPC,due to its rolling optimization feature,can quickly eliminate errors and track the target path.The control path tracking error is less than 10 cm,which verifies the effectiveness and accuracy of the MPC path tracking controller.