Object and ground classification for a mobile robot in urban environment

Abstract

Recently, An interest about the unmanned vehicles is increasing, and a related research has been actively researched. Application systems using the partial element of technologies are commercialized. The information about surrounding environment should be able to use effectively in order to perform a given task such as robot navigation, path planning, and obstacle avoidance. The essential function for a mobile robot is object perception. This paper proposes an algorithm of object detection using stereo camera. The 3D spatial information is obtained by stereo matching algorithm. The reliability of 3D data is defined according to the distance between the object and the camera, and is used in the filtering process. The geometrical features were analyzed by the continuous characteristic of pixels in image. We achieve classification of ground and object.