Cooperative Range-Only SLAM based on Rao-Blackwellized Particle Filter

Abstract

n this paper, a new cooperative range-only simultaneous localization and mapping (RO-SLAM) algorithm is proposed, based on the Rao-Blackwellized particle filter (RBPF). The proposed cooperative approach, which measures ranges between any pair of two nodes cooperatively for the localization, has several advantages over the conventional RBPF based ROSLAM algorithm. Firstly, the particle distribution with the importance weight can be quickly converged to the Gaussian distribution, which yields a fast convergence with the reduced computational burden. Secondly, the proposed algorithm using the inter-node measurements can yield improved localization accuracy. Thirdly, the accuracy is not affected by odometry error unlike the conventional RO-SLAM algorithm because the map estimation is done without moving a robot, and thus the accuracy level is maintained regardless of the sensor network size. These advantages of the proposed algorithm are verified through several real experiments.