Research On Intelligent Vehicle Lane Keeping System For Structured Road

In the development process of more than 100 years from the appearance of automobile to now,it has experienced pure mechanical system stage,automation stage and intelligent stage.The intelligent automobile has great significance in improving safety,alleviating driving fatigue,reducing costs and alleviating traffic jams.Therefore,intelligent vehicle has become a research hotspot in the current academic and industrial circles.With the development of 5g,machine learning,Internet of things and other high-tech development,sensor,computing chip hardware cost reduction and national policy support for intelligent vehicles,the research and investment related to intelligent vehicles presents a "blowout" explosion.The main content of this project is to design a set of perception,planning and control scheme to realize the lowspeed and safe autonomous driving function with lane keeping as the core.The main contents of this paper are as follows:(1)Research on perception of structured road information.Firstly,the lane information of campus environment is analyzed,and the scheme of sensor selection and layout is proposed.Then,the camera internal parameters are obtained by Zhang Zhengyou calibration method,and the external parameters of camera and radar are solved by the corresponding relationship of feature points.After that,image preprocessing,feature detection and lane fitting are used to complete the lane detection of image information.Then,the process of ROI selection,road segmentation,object clustering and border extraction is used to complete the obstacle detection of point cloud information.(2)Research on lane keeping trajectory planning strategy.Firstly,the perceptual information is aligned in space and transformed into matrix information as the input of trajectory planning.Then,considering the complexity and dynamics of campus environment,the complex scene of lane following and keeping is divided into three basic scenes: ideal following and keeping scene,obstacle avoidance scene in the same direction and deceleration scene across obstacles.Then,global path re planning is carried out in the ideal following and keeping scene.In terms of speed planning,an improved trapezoidal speed planning algorithm for dynamic obstacles is proposed based on trapezoidal speed planning.Then,local path planning is carried out in the same direction obstacle avoidance scene,and the velocity is corrected by lateral acceleration.Finally,the research on speed planning is mainly completed because the deceleration scene across obstacles does not involve the change of path.(3)Research on lane keeping motion control strategy.Firstly,the electric control scheme of the longitudinal and transverse control actuator of the car is reformed,and then the control strategies of the longitudinal and transverse control are designed respectively.In the aspect of longitudinal control,considering the influence of road slope on driving,a ramp predictive driving control strategy is established;referring to the way of artificial braking,a fuzzy grading braking control strategy is established;in order to avoid the simultaneous action of driving and braking,the switching rules of driving and braking are formulated.In the aspect of lateral control,in order to eliminate the influence of "time delay" on trajectory tracking accuracy,the lateral control strategy of course prediction is established.(4)Lane keeping strategy simulation and real vehicle verification.Firstly,based on Car Sim and Simulink,the virtual simulation environment and planning control strategy are built respectively,and the joint simulation experiment is carried out.The simulation results show that the car can realize the low-speed and safe autonomous driving function with lane keeping as the core.Finally,the preliminary verification of the real vehicle shows that the car can realize autonomous driving on the simple structured road.