As the reliability of cheap, off-the-shelf autonomous platforms increases, so does the risk posed by intelligent multi-agent systems to military operations. In the contemporary context of the Russo-Ukrainian war alone, we have seen autonomous aerial vehicles and surface vessels deployed both individually and in multitude to deliver critical effects to both sides. While there is a large body of literature on tactical level communications and interactions between agents, the exploration of high-level command and control (C2) structures that will underpin future autonomous multi-agent military operations is a less explored area of research. We propose a quantitative game-theoretic framework to study effective C2 structures in cooperative and competitive multi-agent swarming scenarios. To test our framework, we construct a virtual environment where two adversarial swarms compete to achieve outcomes comparable to real-world scenarios. The framework we present in this paper enables us to quickly test and interrogate different C2 configurations in multi-agent systems to explore C2 as a force multiplier when at a force disadvantage.

翻译：随着廉价商用自主平台可靠性的提升，智能多智能体系统对军事行动构成的威胁也在同步增长。仅在当代俄乌战争的背景下，我们就已观察到自主飞行器与水面舰艇以单体和集群形式被部署，为交战双方带来关键性战术效果。尽管现有文献对智能体间的战术级通信与交互已有广泛研究，但针对支撑未来自主多智能体军事行动的高层级指挥控制（C2）结构的探索仍属研究盲区。本文提出一种基于博弈论的量化框架，用于研究协作与对抗性多智能体蜂群场景中的高效C2结构。为验证该框架，我们构建了虚拟环境，使两个对抗蜂群在近似真实场景的条件下争夺作战效能。本文提出的框架能够快速测试并解析多智能体系统中不同C2结构配置，进而探究在兵力劣势条件下C2作为作战效能倍增器的作用机理。