Real-time SLAM using an RGB-D camera for mobile robots

Abstract

The ability of the localization and the mapping is the essential for the autonomous mobile robot. This is still a popular topic in the robotics community. However most of the SLAM approaches rely on laser scanners that are expensive. With the release of RGB-D cameras, a new class of Visual SLAM approaches became available. The major advantage of the RGB-D cameras is that they provide not only RGB but also the depth information which can estimate the full 6 degrees of freedom of the position in a low-cost way. In this paper, we propose a real-time system that uses an RGB-D camera as the sole sensor to localize the mobile robot and to build a 3D map simultaneously. We evaluated the performance of the proposed system through the datasets and the mobile robot platform. We also compared the proposed system to the FOVIS system. The results show that the proposed system can process at a rate of 25 fps on a CPU core of the command laptop without GPU support while the accuracy is better then 4 cm.