Simultaneous Localization and Mapping Based on PF-MDS

This paper presents an algorithm for the simultaneous localization and mapping (SLAM) problem. Inspired by the basic idea of the fast SLAM which separates the robot pose estimation problem and mapping problem, we use the particle filter (PF) to estimate the pose of individual robot and use the multi-dimensional scaling (MDS), one of the distance mapping method, to find the relative coordinates of landmarks toward the robot. We apply the proposed algorithm to not only the single robot SLAM, but also the multi-robot SLAM. Experimental results demonstrate the effectiveness of the proposed algorithm over the Fast SLAM. The accuracy of the Fast SLAM and that of our proposed SLAM are almost matched with less particles.