Effectiveness of autonomous decision making for unmanned combat aerial vehicles in dogfight engagements

The main objective of this work is to perform a study of the utility of unmanned combat aerial vehicles (UCAVs) in dogfighting (DF) engagements, with DF defined as an aerial battle between two fighter aircraft taking place at close range. The key problem is to assess effectiveness of UCAVs in DF combat when using autonomous decision making based on a representative guidance law and a game-theoretic algorithm. The UCAV DF problem is considered here as a two-player (two fighters), zero-sum, sequential-interaction game with limited information (i.e., each fighter only knows the last three positions of its opponent every time a decision needs to be made). A software simulator has been developed to represent a one-versus-one, clear-sky, close-range aerial battle involving three-dimensional trajectories with high-angle-of-attack (AOA) maneuvers for fighters with similar/dissimilar performance capabilities, considered under four initial conditions: offensive, defensive, neutral, and opposing engagements. Different “levels of intelligence” of the enemy are implemented to validate the performance of the UCAV autonomous decision making against diverse opponents. The simulation-based parametric study elucidates the influence of fighters’ performance capabilities and the fighters’ skill on the outcome of the engagement.