An architecture is proposed which combines artificial intelligence (AI) and game theory to produce near-optimal aerial combat maneuvers. The architecture consists of a flight-path generator, a series of communicating expert objects (CEOs) representing various maneuvers, and an executive. The executive chooses the CEO with the highest score. The chosen CEO passes the formulation of an optimization problem to the flight-patch generator which calculates open-loop controls for the near-optimal flight path, thus closing the loop. The architecture has been realized in the Air Combat Maneuvering Expert (ACME) model, which gives a good representation of one-on-one basic fighter maneuvers.
A hybrid AI/game theory approach to aerial combat simulation
Hybride KI/Spieltheorie-Naeherung zur Luftkampf-Simulation
1988
8 Seiten, 18 Quellen
Conference paper
English
NTIS | 1969
|NTIS | 1985
Coplanar tail-chase aerial combat as a differential game
AIAA | 1977
|A new approach to aerial combat games
TIBKAT | 1970
|