Modeling and Simulation for Design, Development, Testing and Evaluation of Autonomous Multi-Agent Models

U.S. forces are benefiting from automation systems of unprecedented sophistication, empowered by advances in artificial intelligence (AI) and human-systems interaction. In air combat operations, onboard intelligent assistants monitor the aircraft, interpret and carry out commands, and report aircraft and system status, mission progress, threats and alerts. Because pilots and agents are part of a network that may involve ambiguous communications and dynamic roles and responsibilities, Anti-Access/Area Denial (A2AD) raises significant implications for how aircrew can work with their intelligent associates. There is thus a need for training, concepts of operations (CONOPS), and tactics, techniques, and procedures (TTPs) to combat this threat. A2AD effects can have devastating effects on how people and intelligent systems work together. An urgent need has surfaced for a robust capability to simulate, test, predict, and evaluate how these human/agent teams can succeed when faced by both nominal and adverse conditions. In Phase I, we demonstrated a framework that overcomes limitations of current modeling approaches, which are not adequate to properly capture work practices of the socio-technical system. In Phase II, we propose creating a contested airspace simulation testbed that addresses this urgent need using an innovative application of a sophisticated and proven work practice analysis simulation.