An efficient road detection method in noisy urban environment

Road detection is a crucial part of autonomous driving system. Most of the methods proposed nowadays only achieve reliable results in relatively clean environments. In this paper, we combine edge detection with road area extraction to solve this problem. Our method works well even on noisy campus road whose boundaries are blurred with sidewalks and surface is often covered with unbalanced sunlight. First, segmentation is done and the segments which belong to road are chosen and merged. Second, we use Hough transform and a voting method to get the vanishing point. Then, the boundaries are searched according to the road shape. We also employ prediction to make our method achieve better performance in video sequence. Our method is fast enough to meet real-time requirement. Experiments were carried out on the intelligent vehicle SpringRobot on campus roads, which is a good representation of urban environment.