A Real-Time Collaborative Mapping Framework using UGVs and UAVs

Heterogeneous multi-robot systems, combining un-manned ground vehicles (UGVs) and unmanned aerial vehicles (UAVs), offer enhanced efficiency and adaptability in complex environments compared to single-robot systems. This paper introduces a method for constructing high-precision, real-time point cloud maps using air-ground heterogeneous platforms. Utilizing a novel data structure called metascan for basic map building, the method leverages Global Navigation Satellite System(GNSS) data to establish both intra-platform closed-loop and inter-platform co-observation constraints. These are used to create a joint factor graph. By solving the factor graph, the optimized pose data is obtained, thereby realizing the joint mapping of heterogeneous platforms. Field tests conducted in two urban settings confirm the method's ability to accurately and swiftly generate three-dimensional point cloud maps of these areas.