Research On Path Planning And Tracking Of Automatic Parking System Based On Vision And Radar Sensors

The automatic parking system can help people avoid traffic safety accidents caused by improper operation when parking,and it is also an important part of intelligent vehicles,which has been the research focus of major automobile manufacturers and universities at home and abroad.In this paper,we investigate the parking space detection,path planning and path tracking of automatic parking,and use camera and ultrasonic radar to detect parking space for the most common parallel parking scenario,followed by path planning and design a path tracking controller to track the parking path,and verify the effectiveness of the model through joint simulation and intelligent cart test,which provides a part of reference for developing low-cost automatic parking system or assisted It provides a part of reference for the development of low-cost automatic parking system or assisted driving system.Firstly,to address the current situation that the parking space detection of the automatic parking system has a certain degree of leakage and misdetection,this paper adopts two kinds of sensors,camera and ultrasonic radar,to detect the parking space respectively.A series of digital image pre-processing operations are carried out on the parking space images captured by the camera,and the straight lines in the parking space images are detected by using the Hough transform principle,and the program for fitting the parking space lines is written in Visual studio based on the Open CV platform to complete the detection of the parking space.Using the principle of ultrasonic distance measurement,when the vehicle drives parallel to the parking space,the parking space is detected by the relationship between the value of the product of the time of transmitting to receiving ultrasonic waves and the propagation speed of ultrasonic waves and the distance.Secondly,the parallel parking motion process is analyzed based on Ackermann steering principle,and the motion law of the whole vehicle contour is deduced based on the geometric relationship between the outer contour point of the body and the center reference point of the rear axle.After comparing and analyzing the advantages and disadvantages of traditional path planning methods,the B-sample curve with continuous curvature and easy tracking is selected as the parallel parking path in this paper.Combining the environmental constraints of the parallel parking scenario and the kinematic constraints of the vehicle itself,the minimum parking space size required for parking is derived.Then,in order to reduce the effect of vehicle speed on the path tracking effect as well as to ensure that the vehicle meets the vehicle parameter constraints during the path tracking process,the path tracking controller for parallel parking is designed based on the model predictive control theory,and the simulation of the path tracking is carried out in Simulink.Finally,the joint simulation verification of the parallel parking system control strategy is completed by combining the parking scenario and vehicle kinematic model in Carsim and Simulink vehicle motion control model.The intelligent cart test platform was built using a microcontroller,and the control code was loaded in to control the intelligent cart for parallel parking test to verify the effectiveness of the control model.