A Remote RGB-D VSLAM Solution for Low Computational Powered Robots

Abstract

This paper proposes a wireless remote RGB-D visual SLAM solution for low computational powered robots. We develop a ROS implementation of the tools necessary to send RGB-D images over wireless networks. Making use of state-of-the-art compression techniques, such as Run/Variable Length (RVL), a 16-bit depth image compression method, the proposed remote system delivers the same performance as traditional local RGB-D visual SLAM implementations, while also being available to low computational powered robots.