Modeling Autonomous Swarm-on-Swarm Intercept Vehicle Engagements (MASSIVE)

Recent surges in the advancement of artificial intelligence (AI) systems mean that actions that depend on AI are becoming more complex by the day. When this technology is used for the autonomous guidance of cooperative missile swarms operating in the exo-atmosphere and hypersonic regime, the result is missiles that can evade interceptors while maintaining enough energy to continue to their targets. The key to defending against them lies in our own autonomous defense infrastructure and the ability of the AI guiding that architecture to be able to predict and respond to likely evasive maneuvers by the enemy swarm. To address this challenge, Stellar Science and the University of New Mexico (UNM) propose a framework we call Modeling Autonomous Swarm-on-Swarm Intercept Vehicle Engagements, or MASSIVE. This framework will utilize reinforcement learning (RL) informed by differential game theory to guide defensive swarm maneuvers while accounting for enemy evasive movements. It will demonstrate peer-to-peer communication where each swarm member is controlled independently while operating cooperatively. MASSIVE will also provide a testbed and a flexible framework for future development of new swarm behaviors, machine-learning techniques, and objective functions. Approved for Public Release | 24-MDA-11673 (8 Jan 24)