Mobile robot navigation using SURF features

摘要

In this paper, we propose SURF feature based approach for mobile robot navigation. In this approach, we capture environmental images by using omni-directional camera and extract features of these images by using SURF. We treat these features as landmarks and estimate robot self-location and direction of motion. SURF features are invariant under scale change and rotation and robust under image noise, change in light condition and change view point. Therefore, SURF features are appropriate ones for robot self-location estimation and navigation.