Visual navigation: detection of the navigable free space

The general context of this work is the guidance of an autonomous mobile robot directly from the information contained in a sequence of images. Because of the complexity of the problem to be solved, the authors decided first to concentrate on the implementation of a behaviour of lower level rather than route planning or environment exploration: the avoidance of obstacles. This article deals with the identification of image regions corresponding to the navigable free space. It presents a method of obstacle detection based on preliminary determination of the parts of the image corresponding to the ground over which the vehicle moves