Simultaneous Localization and Mapping for Mobile Robots Using Stereo Vision Camera

Simultaneous Localization and Mapping (SLAM) technology is widely used in the field of service robots and unmanned systems. ORB-SLAM2 is one of the famous feature point based real-time SLAM systems, which has high positioning accuracy, robustness, and the ability to operate in different environments. However, the entire SLAM system can only estimate sparse mapping. In order to construct the 3D point cloud map, the color information in the image needs to be transformed and combined with the depth information captured by the stereo camera. In this paper, we construct a binocular SLAM framework called MYNT-vSLAM to improve the sparse mapping of ORB-SLAM2. Experimental results verified the feasibility of this algorithm, which is robust and fast in complex indoor environment, and can be applied to location and 3D point cloud map construction.