A visual localization method for mobile robot based on background subtraction and optical flow tracking

Abstract

In this paper, a newly vision-based localization method for mobile robots is presented. The cost of hardware for this localization system is very low. Based on background subtraction and optical flow tracking, the presented method overcomes the shortcoming of general background subtracting method- being sensitive to noises and interferences of other mobile objects. Besides, the presented method is enough robust to find the new optical flow point after it is lost. Experiments reveal that this presented method is effective and is able to meet needs of localization for mobile robots.