Research Of Vehicle Lateral Control Strategy Based On Machine Vision

With the continuous improvement of economic and living standards, the number of cars is growing rapidly, and the traffic accidents is also a rapid rise. In order to reduce the traffic accident, states researchers are actively exploring smart car technology. On the basis of a smart car technology at home and abroad to put forward the vehicle lateral control technology based on machine vision, this technology has four parts:binocular camera calibration technology, path detection technology,3D reconstruction and vehicle positioning, vehicle lateral control.Binocular camera calibration part first uses two CCD cameras to form a binocular stereo vision system, two kinds of camera models are established, which are linear and distortion, the advantages and disadvantages of two kinds of camera calibration techniques are analyzed. Then,the calibration method based on plane template is selected to obtain the internal and external parameters of the binocular stereo camera, and the analysis of the parameters is accurate.In road detection part, firstly, the preprocessing of the gray scale, filtering, adaptive threshold segmentation is carried out to remove the noise in the road image and the region uninterest, then edge detection algorithm is used to extract the edge information of the road, then, use the corrosion expansion and other morphological processing to fill small holes. Finally, the Hough transform is used to fit the navigation path successfully.In 3D reconstruction and vehicle positioning part, firstly, the left and right road images are corrected, which makes the area of the distorted image to the minimum, using the based on SIFT feature matching and region matching to find the corresponding point, using SIFT feature matching to successfully match and get around two images of the mosaic map, region matching gets the disparity map of the left and right two images. Based on the disparity map, the internal and external parameters of the camera are used to carry out the three-dimensional reconstruction of the road image. The relative position and attitude between the vehicle and the road can be calculated by using the detected navigation path.In the vehicle lateral control system, the mathematical model of vehicle lateral control based on preview mechanism is established, then the fuzzy sliding mode controller based on equivalent control is designed for the chattering problem of sliding mode controller. Using the fuzzy algorithm to smooth the control signal so as to reduce the chattering amplitude. The simulation results show that the lateral deviation and lateral deviation angle can converge to zero in a short time, and the system chattering amplitude can be effectively reduced, which is more accurate for the desired path tracking.In this paper, a large number of experiments to analyze and verify the effectiveness of the above four technology modules, the analysis results show that the parameters of binocular camera calibration module are accurate, the path detection module can accurately detect the navigation path, three dimensional reconstruction and vehicle positioning module can obtain the relative position in real time, and the fuzzy sliding mode controller based on equivalent control in the vehicle lateral control module can effectively control the vehicle transverse through the experimental simulation.