Reserch On Lateral Motion Control Of Autonomous Vehicle In Lane Changing Process Based On Vision

With the rapid increase of car ownership,traffic accidents and environmental pollution are becoming more and more serious and need to be resolved.And the rapid development of electronic information technology will promote the transformation of the traditional automobile industry and promote the transformation of automobiles from travel vehicles to large intelligent mobile platforms.This issue comes from a school-enterprise cooperation of autonomous vehicle research and development project in Huadu District,Guangzhou.The focus of research is the lateral motion control of autonomous vehicle in specific scenes based on vision.The driving state on the road mainly includes driving with the car and changing lane.The lane changing process is complicated,which is a typical application problem of the autonomous vehicle;The lateral motion control of autonomous vehicle is to track the desired trajectory by controlling the front wheel angle,which is a core research problem of autonomous vehicle motion control.Therefore,this paper focuses on the research of lateral motion control of autonomous vehicle in lane changing process based on vision as the research content.Firstly,aiming at the problems existing in the YOLO target detection algorithm,this paper studies the improved target detection algorithm based on YOLO,and designs the vehicle detection model based on YOLO improved algorithm for the needs of the issue.Then,the monocular vision front distance estimation model based on geometric imaging method is derived.The two-part algorithm is combined to realize the real-time detection and distance estimation of the vehicle ahead.Secondly,the vehicle lane changing process is analyzed,and the lane changing strategy based on the safety distance model is designed.Compared with the common lane-changing trajectory planning method,the state quantities of various lane-changing trajectories are analyzed,and the lane-changing trajectory planning is realized by the fifth-order polynomial.Then,based on the vehicle monorail model and the magic formula tire model,the vehicle dynamics model under the small angle hypothesis is established.And a lateral motion controller based on linear time-varying model predictive control is designed.In the Simulink/CarSim cosimulation environment,the performance of the designed lateral motion controller under different design parameters,different road surface adhesion conditions and different vehicle speeds was verified.Finally,for a pure electric vehicle platform,design software and hardware system modification program.The real vehicle experiment proves that the lateral motion control system of autonomous vehicle in lane changing process based on vision can basically estimate the front distance in real time and control the vehicle to change lanes smoothly.