V-SLAM: Vision-based simultaneous localization and map building for an autonomous mobile robot

Abstract

Simultaneous localization and map building (SLAM) is a desired feature for autonomous mobile robots. This SLAM approach allows the robot to create a map on the fly and then backtrack to further explore the area without human interaction. Data from the robot's encoder and sonar sensors are used along with depth information from a stereo camera vision system to explore and map the surroundings. The Speeded Up Robust Features algorithm (SURF) is used to visually identify landmarks in the environment.