Future battlefields are expected to be heavily contested by near peer adversaries boasting integrated air defense systems (IADS) with advanced radar systems, lethal surface to air threats, and effective electronic warfare (EW) capabilities. The collaborative nature of munition swarms makes them a potentially effective technology for not only delivering kinetic effects to the IADS components, but also illuminating and sensing enemy technology and tactics. The hypothesized multi-wave chain of collaborative weapons could be hugely effective, but significant research is needed to understand how best to deliver the desired multi-wave capability. In Phase I, Toyons proposes the research and development of collaborative munition swarming behavior using Multi-wave Adaptation through Phase Learning (MAPL). Toyon will perform concept development with the Air Force for a reference mission of a multi-day campaign. Toyon will leverage an existing single-day counter IADS scenario developed in prior SBIR Phase II programs as a departure point. Toyon will develop a forensics pipeline to aid in Mission Data analysis and iterative retraining of a reinforcement learning (RL) agent model against the multi-day scenario to answer questions of what data is most important for communication and logging, how to rapidly update algorithms, and how to select algorithms.
