Multirobot Collaborative SLAM Based on Novel Descriptor With LiDAR Remote Sensing

Abstract

Geospatial data is essential for urban planning and environmental sustainability. Utilizing multiple robots, each equipped with 3-D LiDAR for remote sensing, to collaboratively construct environmental maps can significantly enhance the efficiency of geospatial data collection. However, efficiently identifying overlapping areas between robots and accurately merging the maps constructed by different robots remains a pressing challenge. This study proposes a multirobot collaborative simultaneous localization and mapping (SLAM) method based on a novel environmental feature descriptor to address this problem. In this method, a distributed multirobot collaborative SLAM system is first constructed. Then, an SLAM algorithm that integrates intensity features and ground constraint is proposed for the robots in the multirobot SLAM system. Additionally, a multilayer hybrid context descriptor is introduced to detect overlapping areas between different robots. To validate the effectiveness and advantages of our method, we conducted benchmark comparisons with other approaches. Our multirobot collaborative SLAM method demonstrated favorable experimental results.