Improvement of map fusion algorithm for multi-robot simultaneous localization and mapping

This paper mainly studies the real-time fusion of the submaps of multi-robot simultaneous localization and mapping (SLAM). In this paper, an improved HybridSLAM algorithm is proposed, which can observe and update multiple landmarks at the same time. The improved algorithm uses the most precise observation among multiple landmarks to correct the robot pose according to the algorithm of FastSLAM2.0. Then, based on the improved HybridSLAM algorithm, this paper further proposes a multi-robot improved HybridSLAM method (MR-IHybridSLAM). Each robot runs the IHybrid- SLAM algorithm to build a submap at different initial positions, and sends the submap information to the same workstation in real time. According to the Kalman filter (KF) principle, submaps built by each robot individually are merged into a global map. Finally, the fused map of multi-robot is constructed by simulation experiments, and subsequently the map errors of the fused map and other maps constructed by single robot FastSLAM and HybridSLAM algorithm are compared to verify the accuracy, fastness and feasibility of the algorithm. © 2019, Editorial Department of Control Theory & Applications South China University of Technology. All right reserved.