Researchers have developed a new approach to generating game-theoretically optimized courses of action for a Blue Unmanned Aerial System (UAS) swarm against an adaptive Red adversary in a communication-degraded environment. The method, called UC-PSRO, combines three mechanisms: policy-space response oracles with self-play, utility-conditioning of the Blue policy, and a curriculum annealing communication-graph edge dropout during training. The authors evaluate their approach on a synthetic scenario and find that the communication-dropout curriculum alone provides the strongest mission-completion rates, improving as denial increases. However, adding utility-conditioning and PSRO self-play slows convergence within a fixed budget, with no reliable exploitability advantage over a fixed-opponent policy. --- Why it matters: This research matters to engineers working on autonomous systems because it explores how to optimize decision-making in complex, dynamic environments where communication is limited or unreliable. The results have implications for the development of robust and efficient control strategies for swarms of UAS. Source: https://arxiv.org/abs/2608.15372