Design of simulation-based pilot training systems using machine learning agents

Abstract The high operational cost of aircraft, limited availability air space, and strict safety regulations make training fighter pilots increasingly challenging. By integrating Live, Virtual, Constructive simulation resources, efficiency effectiveness can be improved. In particular, if constructive simulations, which provide synthetic agents operating vehicles, were used to a higher degree, complex scenarios could realised at low cost, the need for support personnel reduced, this work, inspired by recent improvements techniques artificial intelligence, we take user perspective investigate how intelligent, learning help build future systems. Through domain analysis, study, practical experiments, identify important agent capabilities characteristics, then discuss design approaches solution concepts systems utilise improved value.