Multi-agent-based battlefield reconnaissance simulation by novel task decompositionand allocation

摘要

In order to support the advanced concept technology demonstration of intelligence reconnaissance activities, multi-agent-based battlefield reconnaissance simulation is studied to provide enough detail to examine dynamic tactical process by a novel method of task decomposition and allocation. Our work includes four parts: (1) describing task by means of Behavior Definition Frame; (2) analyzing task by presenting a role-agent mapping mechanism, and thus setting up a mapping from a role to an agent, i.e., from battlefield reconnaissance system's real components, to entity agents; (3) decomposing task based on logistic relationships between real components in battlefield reconnaissance system organization, by using a Task Tree decomposition approach, and allocating task by using task allocation processing module; (4) proposing case study and implementing simulation. The result demonstrated in the case study proves the feasibility and efficiency of our method and model.