Environmental mapping for mobile robot by tracking SIFT feature Points using trinocular vision

Abstract

This paper presents a SIFT based map building and self localization method for mobile robots. First a mapping robot with trinocular vision builds the 3-D keypoint map of unknown environment with a high accuracy and then the working robot with a monocular vision localize the own position by matching the SIFT keypoints to the map. Experimental results of mapping and localization for a real indoor scene is shown.