DSLAM: Towards a real-time robot mapping approach

Abstract

This paper presents a Distributed Simultaneous Localization and Mapping (DSLAM) technique to realize Laser Range Finder (LRF) and stereo vision based 3D robotics fast mapping. The motivation stems from the fact that the optimization procedure of laser scan matching involved in robot SLAM and the stereo ranging is both computationally intensive. In DSLAM, parallelism mechanism is studied by using multi-thread architecture. Experimental results demonstrated the effectiveness of DSLAM in improving the efficiency of robot mapping.