A Hierarchical Planning Strategy for Complex Unmanned System

The research and widespread application of unmanned systems are profoundly impacting human production and lifestyle. Among unmanned system research, unmanned equipment is typically modeled as agents, and multi-agent collaborative task planning is one of the core issues in the practical application. Particularly in situations where the task-agent combinations and the requirements for agent assignments are uncertain, the interplay of various task elements and constraints can lead to combinatorial explosions. Traditional mixed-integer optimization and temporal logic methods struggle to efficiently complete task assignments, while data-driven approaches face challenges related to data scarcity and reliability. This paper focuses on the multi-agent collaborative task planning issue, primarily addressing two difficulties: the optimization of collaborative task-agent combination requirements and the optimization of agent execution strategies. By decomposing the multi-level combination optimization problem, we design intelligent algorithms for rapid optimization and solving. Combining the speed advantages of data-driven methods with the interpretability benefits of formal logic approaches, the proposed method can complete large-scale agent collaborative task assignments within seconds.