Research On Automatic Parking Algorithms Based On Single Point Preview Deviation Model

With the improvement of people's living standards and the rapid development of automobile industry,the car ownership in our country is increasing gradually.However,the growth rate of automobiles is much faster than the speed of building parking spaces,which leads to the gradual reduction of parking space and many traffic accidents are caused by parking errors.Therefore,automatic parking system has become a research hotspot of domestic and foreign automobile enterprises and scientific research institutions.Under the background of the rapid development of intelligent driving technology,this paper proposes a single-point preview deviation model to solve the problem of high constraints on starting position of automatic parking system,and to achieve a variety of parking modes.Firstly,the environment model and vehicle kinematics model of automatic parking are built,and the vehicle initial position and posture are deduced according to the global positioning scheme and the local positioning scheme based on parking line recognition.A single-point preview deviation model is designed to plan parallel parking and vertical parking paths,and the model is divided into fixed-point preview and rolling preview according to the state of preview points.The position and posture constraints of starting point,transition point and target point of vehicle in parking environment are determined,and the influence of pre-aiming point and the shortest pre-aiming distance on the vehicle storage accuracy is analyzed.The model parameters are determined by minimizing the lateral displacement error and heading angle error when the vehicle reaches the target position.The simulation environment of parallel parking and vertical parking is built in MATLAB/Simulink.The results show that the single-point preview deviation model can plan various parking modes and reduce the constraints of the automatic parking system on the starting position.In this paper,three parking modes are planned in vertical parking environment by using single-point preview deviation model,and the parking mode decision module is designed according to the heading angle of the vehicle at the entrance of the parking space,so that the optimal parking mode can be chosen when the vehicle is near the parking space,which can not only reduce the parking space,but also improve the parking accuracy.Finally,an experimental vehicle based on LEGO EV3 controller is built to validate the single-point preview deviation model.The software architecture of the vehicle is hierarchical.The underlying driver is designed in MATLAB/Simulink and compiled into the controller.The control strategy is compiled in Python,and the communication between the vehicle and the computer is realized by using TCP/IP protocol.Parking experiments are carried out in parallel and vertical parking scene respectively,and the results show that the single point preview deviation model can be well applied in automatic parking system.