Scanning-based floor region identification from omnidirectional images for mobile robot localization

For a mobile robot, it is necessary to have the ability to move accurately and autonomously in an unknown environment. The identification of neighboring floor region is very important for robot localization. In this paper, we propose a method imitating the way of laser range finder for detecting the closest objects with the Canny edge detection. The identification of moveable floor region is important for the localization and navigation of mobile robots. The proposed approach has been experimentally tested in various indoor environments such as library and corridor.