Mobile robot localization with sparse landmarks.

Abstract

This paper describes a mobile robot system designed to explore and map an area such as is encountered in urban search and rescue simulations. The robot uses homogeneous artificial landmarks for localization as it builds metric and topological maps, determining landmark distance and bearing with groundplane calculations from a single camera and using Kalman filtering techniques to perform localization. When implemented on a Magellan II mobile robot, the localization technique correctly localized the robot while exploring and mapping.