Tactical Pursuit Point-Based Autonomous Decision Making for Dogfights in Defensive Situations

Although extensive research has focused on autonomous decision-making (ADM) for offensive air combat, studies addressing adverse conditions remain an open issue. This paper focuses on the defensive maneuver decision-making problem in close-range air combat. A defensive ADM system based on the tactical pursuit points (TPPs) approach is studied, and four defensive tactical pursuit bases that have distinct defensive tactical significance are proposed to establish the decision space. Consequently, the defensive autonomous air combat problem can be represented as an optimization problem involving a four-dimensional decision vector within this space, simplifying the complexity while maintaining tactical effectiveness. Additionally, a reinforcement learning-based reward mechanism is applied to obtain the decision vector, which combines the tactical pursuit bases (TPBs) to generate TPPs that guide the fighter. Examples are provided to illustrate the outcomes of this study.