Abstract
This paper presents a method for autonomous mobile robots to recognize sidewalks and paths using stereovision based system. The 3D and color information are converted to an elevation map which will reduce the data size significantly compared with point clouds. The height and color of the grids in the elevation map are used to calculate the score map, which include height information among with its gradient and the road classification information. The road recognition is performed by making a histogram from sampled area and comparing other grids with the histogram. The grid with the color information that is included in the histogram will be recognized as the road area.
