Multi-platform Cooperative Target Disposal Method Based on Dynamic Spatiotemporal Window and Multi-Agent Path Planning

Abstract

Aiming at the problems of overlapping explosion hazard zones and large task waiting loss in the traditional batch mode of multi-platform delayed blasting cooperative operations in obstacle clearance and regional target disposal scenarios, this paper proposes a new path solving method integrating the idea of Multi-Agent Path Finding (MAPF). First, a spatiotemporal cylindrical model of explosion consisting of three stages: delayed detonation, hazard persistence, and safety recovery is established. The explosion hazard zone is modeled as a dynamic obstacle with a definite time window to characterize the time-varying characteristics of explosion hazards. On this basis, a multiple traveling salesman target assignment model with spatiotemporal window constraints is constructed, introducing coordination constraints such as safety of consecutive targets for the same platform, simultaneous disposal of multiple platforms, and a minimum waiting time insertion strategy. An improved Non-dominated Sorting Genetic Algorithm II (NSGA-II) is used to generate a Pareto optimal allocation scheme that balances task completion time and total voyage. For path planning, a hierarchical MAPF algorithm combining Conflict-Based Search (CBS) and spatiotemporal A* with hovering action is proposed to achieve the global optimal trade-off between detour and waiting costs. Simulation experiments show that the multi-platform cooperative algorithm based on dynamic spatiotemporal window and multi-agent path planning has higher reliability and safety when performing target disposal tasks.