Multi-map Collaborative Exploration toward Large-Scale 3D Mapping Using UAV *

In order to reduce the maintenance cost of map and enhance the search efficiency in autonomous exploration of 3D environment, this paper proposed a novel algorithm toward large-scale 3D mapping for aerial robotic platforms. The proposed planner employs a multi-map collaborative scheme, in which the accurate boundary detection is performed on a high-precision submap with limited range, and the low-cost global map supply the information of submap to guide the rough direction of exploration. As a result, the frontier is incrementally searched instead of repeatedly searching the entire known map and the optimal frontier is decided. In 3D environment, its performance is evaluated by comparing with Wavefront Frontier (WFD) algorithm and Next Best View Planner (NBVP) algorithm. The proposed method reduces more than 20% and 60% of exploration time respectively.