A multi-unmanned aerial vehicle fast path-planning method based on non-rigid hierarchical discrete grid voxel environment modeling

•An undirected graph environment modeling method based on a non-rigid hierarchical discrete grid structure was proposed.•The undirected graph modeling method was capable of improving identification ability and unifying conflict description.•A fast multi-UAV path planning algorithm named CBOP was proposed for large-scale (>100 UAVs) problems.•CBOP took the whole UAV as the resolution unit in the high-level solver and use priority calculation to realize rapid path planning. Multi-unmanned aerial vehicle (multi-UAV) path planning involves determining no-conflicting paths for multiple UAVs. Given the shortcomings of existing planning algorithms, such as number limitations and low efficiency, we studied an undirected graph environment construction method based on a non-rigid grid data model to unify conflict descriptions and simplify the conflict detection process. Accordingly, we propose Conflict-based Objective-oriented Prioritization (CBOP), a fast multi-UAV path-planning algorithm. CBOP takes the whole UAV as the resolution unit in the high-level solver and uses priority calculation to avoid massive branch calculations. Experiments show that: (1) Compared with the voxel environment constructed by the rigid grid data model, adopting our undirected graph can reduce the planning time by 6.7% and the planned path length by 7.4% on average when the number of UAVs exceeds 100. (2) Compared with the Conflict-based Search with heuristic (CBSH), one of the most effective variants of CBS, the CBOP is more suitable for solving large-scale problems. The efficiency can be increased by 35.1 times on average when the number of UAVs is between 100 and 500.