Research On Automotive Lane Keeping Assist System Under Virtual Driving Environment

With rapid increase of vehicle possession, the number of traffic accidents becomesincreasingly high. While many factors contribute to the growth of the accidents, includingvehicle defect, traffic jam, and drivers’fault, to name a few, It is no doubt that human drivershave been the major cause to the majority of the accidents, due to distraction, illness,tiredness, operational mistakes, etc. Therefore, research and development of advanced driverassistance systems (ADAS), including driver assistance systems and active safety systems,and even autonomous driving vehicles are critically important. Lane keeping assist system(LKA) is one of the key systems.Lane keeping assist system detects lane departure by visual sensors and adds steeringresistance through electric power steering system (EPS) in order to help driver from lanedeparture unconsciously. On one hand, because of the low cost, coupled with thedevelopment of image processing and pattern recognition technology, in recent yearscameras on board and vision-based auto intelligent technology develop rapidly. One theother hand, with the rising popularity of electric power steering system, steering controltechnology based on EPS matures increasingly. So lane keeping assist system which is basedon visual sensor and EPS torque control becomes popular naturally.Currently tests on field are still the main platform for auto intelligent technology andproduct development, testing and verification. But field tests are of high cost, long period,poor repeatability, which can’t reflect the change and diversity of the working condition andtend to have high risk. Although simulation technology is becoming an important technicalmethod of automobile product development, the existing simulation platforms focus on theperformance of the vehicle more than the simulation of the driving environment,environment sensing and human machine interaction. So they are hard to meet the requirements for lane keeping assist system development.This paper proposes a technology and method of developing a lane keeping assistsystem under virtual driving environment which includes a3D digital virtual experimentalenvironment created based on computer graphics, road and lane models, a camera model,and a steering simulation system with torque feedback. Under this virtual environment, wecan set different experimental conditions according to different requirements and get theimages by a camera model. On one hand, we can set various experimental conditions whichmay influence on camera detection, including the weather, light, roads and lane lines, etc.when the experimental conditions are changed, what we need do is to set in computer simply,and drivers can complete different experiments in the lab without moving to differentexperiment sites. This will greatly reduce the consumption of both cost and time on settingup and replacing the experiment sites. On the other hand, test under virtual environment canensure repeat experiments under same experimental conditions and to guarantee the personalsafety.The thesis is organized as following:In chapter1, the subject background and the significance of the research are discussed.The traditional method of research based on field test cannot meet the needs of autointelligent. While the research based on virtual environment is more efficient. And the maincontent in this paper is given.In chapter2, according to the requirements of research on Lane Keeping technology,the function architecture, hardware architecture and communication architecture of virtualdriving environment are given.In chapter3, in order to simulate the process of real camera getting photos, a model ofcamera was established. Through comparing photos obtained by real camera with by modelof camera, verify the validity of virtual driving environment.In chapter4, a steering torque feedback model is given and the influence factors ofsteering force are analysed firstly. Then through comparing the simulating result of the model with the field test data, verify the validity of the steering force model.In chapter5, the overall architecture of the research on lane keeping assisting systembased on virtual environment is given, including lanes detection under environment, driverintention recognition and lane keeping torque control strategy. And details will be introducedfor each part.In chapter6, a virtual simulation environment is build, and the lane keeping system istested based on this environment.In chapter7, summarize the whole research contents, and the further research isdiscussed.