Mobile robot localization using a light-stripe sensor

This paper presents an absolute localization system for a mobile robot moving on a flat ground, in a structured environment. The method makes use of 3D images provided by a light-stripe sensor. The images represent the sections in a horizontal plane of characteristic edges of the environment, whose model is known at the very outset. The goal is to extract at least two identifiable points out of the observed laser stripe and to determine the position and the heading of the robot considering a frame fixed at the observed edge. The experimental results obtained with an indoor mobile robot show that the localization errors are of the order of some centimeters in location and one degree in heading.