Robust pose estimation and tracking system for a mobile robot using a panoramic camera

A localisation system is an essential knowledge for a mobile robot to be able to freely navigate in its world. In this paper, pose estimation and tracking of a mobile robot is presented for an indoor cluttered environment using only an overhead panoramic vision system. The method presented is fast without requiring unwrapping of the panoramic view. It is assumed that the robot's workspace is 2D planer surface. A system combining mean-shift, Kalman Filter and Hough Transform based tracking is used to improve the result. Experiments show that the method is capable of robustly localising and tracking the robot in cluttered scenes even with variations of illumination and brief periods of occlusion.