The Air Force, in its 2030 Science and Technology Strategy, emphasizes the need for a collaborative and autonomous network of systems and effects, and distributed sensing across domains. Distributed, collaborative systems require complex control algorithms. Introducing autonomy requires formal guarantees of system safety and correctness in order to deploy in an assured, trustworthy fashion. Constructing such controllers, however, poses significant challenges both from a verification and a computational perspective. Verus Research proposes to integrate technologies developed by UT Austin that have been demonstrated to provide provably correct and safe control strategies for teams of agents with Verus Researchs existing capabilities in autonomous satellite verification and control design, developed for AFRL. We will specifically utilize a hierarchical framework that decomposes the multi-agent control design into simpler sub-problems at the mission, network, and agent levels. The methodology we propose for synthesizing trusted, autonomous multi-agent control strategies has several advantages, including (1) incorporation of requirement satisfaction into the control design, leading to (2) guarantees of safety and performance, and (3) reduced computational demands, which are a known hurdle to trusted autonomy.
