Research On Parking Space Selection And Automatic Parking For Intelligent Vehicles

Parking is one of the most complicated yet difficult driving maneuvers,in particular,to many amateur drivers.Therefore,intelligent technology can be integrated into the vehicle control to realize the automatic parking function.That frees humans from tedious driving tasks and improves driving safety,which has broad market prospects and practical significance.At present,research on automatic parking has become a hotspot in the industry and has achieved staged results,but there are still limitations in the research of some functions that need to be resolved.Parking space selection is an important part of automatic parking fuction,the current research mainly relies on the induced allocation strategy of parking lots.This strategy does not take into account the dynamic obstacle information during the parking process,so it cannot adapt to complex and dynamic parking lots.In the study of berth allocation rules,the existing studies have not performed a rational analysis on the selected attributes,It is very likely that there is a large correlation between the selected attributes,which not only complicates the calculation,but also increase the probability of superimposing weights.That will make the berth allocation result unreasonable.Therefore,a parking space selection model with vehicles as the main body was proposed.This article assumes that vehicles use network information,map information,and on-board camera and radar detection information as the basis for berth selection,and extracts several factors that affect berth selection.Aiming at the problem that the parking lot's induced allocation strategy cannot consider the dynamic obstacle information,this paper uses a multi-arc pre-planning method to convert the impact of obstacles on the difficulty of parking into the size of the trajectory safety region,and finally converted into the corner margin of the front wheels according to the Ackerman steering,which was considered as a new attribute into the parking space selection.For the selected decision attribute,this article first performs a factor analysis on the attribute information,excluding the correlation between the attributes,and ensuring the accuracy of subsequent decisions.Based on the entropy weight method,the data value reflected by each attribute information is mined,and assign weights to each attribute based on this;When evaluating the merits of each berth,the best berth is determined based on the "ideal point" solution combined whith the weight information;Finally,this paper designs a typical parking lot scenario and verifies the the model's rationality for attribute weight allocation,and the berth optimization results meet expectations.For the research of parking motion control,most of the researches regard parking as a stable low-speed process.Based on this assumption,the vehicle model is simplified to rigid body motion,ignoring the vehicle's dynamic response and lateral sliding.Good results were obtained when the vehicle was stable at low speeds,but the adaptability to speed was weak.The actual speed of reversing and entering the warehouse is not stable,and there is a change in speed.On the other hand,under the spacious conditions that allow,especially after the popularization of intelligent technology in the future,humans will participate less in the parking environment.Increasing the speed to improve parking efficiency,but if the speed is slightly increased,the dynamic response of the vehicle cannot be ignored.Therefore,the simple rigid body assumption cannot well adapt to changing vehicle speeds,nor can it provide convenience for vehicle speed control during parking.Therefore,this paper re-establishes the parking dynamics model.For the case where the parking control of vehicles was limited to low speed in the past,this article attempts to increase the vehicle speed and amplify the dynamic response of the vehicle.Firstly,by analogy of forward driving,a reverse monorail model is established to fully compare the difference of vehicle dynamics during forward and reverse driving.The conclusion of the instability of the system when reversing is obtained through analysis;the yaw angular velocity and lateral velocity of the vehicle are fed back to the front wheel rotation angle using the state feedback method,and the closed-loop pole of the system is corrected to the negative half-axis,which changes the stability of the vehicle;The closed-loop pole adjustment takes into account the system's response speed and vehicle steering structure limitations,ultimately making the system not only stable,but also closer to the motion characteristics of the actual vehicle when reversing,ensuring the premise that the system must be stable before vehicle control.Finally,a model predictive control method based on the reversing dynamics model is established.In this paper,the reversing dynamics model is brought into the model predictive control,which fully considers the dynamic characteristics of the system;The discrete reversing dynamics model at different vehicle speeds improves the adaptability of the system control to the vehicle speed;Based on the robustness and peace of the model predictive control Compliance,the system can still ensure better control results after increasing the vehicle speed.