Research On Lane Keeping Control Algorithm Based On Machine Vision

In recent years,intelligent vehicle has gradually become the focus.It represent the development trend and its development constantly innovates related theories and technologies.Lane keeping assistance system is an important part of intelligent vehicle,which integrates many technologies such as computer vision,image processing and control theory.This paper studies the environment perception and control algorithms in lane keeping assistance system separately.The LKA first needs to extract the relevant information of the lane in road images.A multi-lane lane detection algorithm is proposed.Firstly,the edge of the lane is extracted by using the gradient information and the width of the lane on the grayscale image.Then the Hough transform is used to detect the lines according to the edge points in the image partition.Finally,the control points are determined according to the partition lines and fits lane lines using cubic uniform B-spline curves.Particularly,using the vanishing line of the lane to reverse the lane marking of the dotted lane enables the algorithm to effectively detect a variety of lanes;the edge points extracted on the gray level graph enhances the algorithm robustness.After a variety of experimental verification,the algorithm can accurately and quickly detect a variety of lanes.After the lane detection,this paper also introduced the conversion method of the pixel coordinate system to the world coordinate system in order to use the lane information and vehicle location information for the subsequent lateral control algorithm.For lane keeping control strategy,this paper firstly builds a seven-degree-of-freedom vehicle model based on the magic tire model.Then,based on the road information,vehicle location and status information,TLC model is established to determine the lane departure.Finally,a lateral controller is established based on the single-point pre-cursor optimal curvature driver model.In order to verify the feasibility of lane keeping control strategy,a simulation was carried out with CarSim/Simulink.According to different target paths and vehicle speed,four simulation conditions were set up.Then the four simulation conditions are analyzed respectively.By analyzing the comparison chart between the actual driving path and the target path,the lateral acceleration curve and the turning angle curve of the vehicle,it shows that the lane keeping control strategy designed in this paper can effectively control the car to follow the target path.