UAV Interception and Confrontation Maneuver Decision-Making Based on Reinforcement Learning

Abstract

In recent years, artificial intelligence technology has performed outstandingly in game confrontation tasks. Based on the characteristics of UAV interception and confrontation process, this paper constructs an interception maneuver strategy learning and training environment based on reinforcement learning methods, including UAV model construction, maneuver decision-making space construction, reward and punishment signal design, enemy UAV strategy design. In order to effectively improve the exploration efficiency of the algorithm, this paper uses expert knowledge as heuristic information and proposes an improved heuristic strategy to avoid initial blind exploration while retaining the optimization ability of the greedy strategy. And completed the simulation verification under the set three-dimensional scene.