Metric localization using a single artificial landmark for indoor mobile robots

Abstract

We present an accurate metric localization method using a simple artificial landmark for the navigation of indoor mobile robots. The proposed landmark model is designed to have a three-dimensional, multi-colored structure and the projective distortion of the structure encodes the distance and heading of the robot with respect to the landmark. Catadioptric vision is adopted for the robust and easier acquisition of the bearing measurements for the landmark. We propose a practical EKF based self-localization method that uses a single artificial landmark and runs in real time.