Self-localization of a mobile robot without camera calibration using projective invariants

In this paper, we propose a visual-based self-localization algorithm of an indoor mobile robot. The algorithm does not require calibration and can be worked with only a single image by using the projective invariant relationship between natural landmarks. The position of the robot is determined by relative positioning. Also, the method does not require previous information of the position of the robot. The robustness and feasibility of our algorithm have been demonstrated through experiments in hallway environments. Ó 2000 Elsevier Science B.V. All rights reserved.