Research On Lane Keeping Algorithm Of Intelligent Vehicle Based On Vision Navigation

There are some kinds of airport enclosures road, for example, dirt road, cement road and asphalt road, and the environment on both sides of the road is also complicated. There are some navigation ways for intelligent vehicle at the airport enclosures road. The navigation ways based on the vision has the advantage of the low cost and rich environmental Information. This article focuses on the visual navigation methods of the intelligent vehicle at the airport enclosures road environment. The main contents are as follows:First, the segmentation algorithm for airport enclosures road image is designed through the analysis of the road image processing method. To enhance the speed of image filtering, an improved fast median filtering algorithm is researched; then in order to enhance the robustness of the road boundary edge detection, the improved Sobel operator is used to highlight the edge of the road boundary information of the image and suppress the interference of edge information. Through experimental comparison of common area segmentation algorithm, the region growing algorithm that can effectively overcome the water stains and shadow interference on the road has stronger anti-interference.Secondly, for the complexity of the airport enclosures road environment, the method that fuses edge detection and adaptive region growing is proposed. The robustness of the road segmentation algorithm is improved by the adaptive change of the guideline of region growing algorithm. According to the characteristics of different regions of the image the partition processing method is taken to improve the speed of the algorithm. The fusion of edge and regional features make extraction road borders more precise.Again, for the lane keeping problem, the control method of PID and fuzzy control rules based on the preview distance is studied. The point in front of the vehicle is selected. The distance of preview point to the target path takes the PID control, and the steering rate takes fuzzy control. Then according to the case of one or two road boundary detection, take the appropriate treatment, and apply the results to the control algorithm. This control method simulates the process of manual driving the car. This is a simple and efficient method.Finally, the experimental results show that the algorithm of the road boundary detection and lane keeping algorithm proposed in this paper can satisfy the need of the intelligent vehicle vision navigation at airport enclosure road environment.