Hierarchical Decision-Making Framework for Multiple UCAVs Autonomous Confrontation

Abstract

Autonomous decision-making for air confrontation between unmanned combat aerial vehicles remains hard to be designed due to dynamic situations and complex interactions. Rule-based decision-making methods provide a powerful solution with better interpretability. However, various hand-crafted rules may result in conflicts and poor scalability issues. To overcome this problem, this work proposes a hierarchical decision-making framework called State-Event-Condition-Action (SECA), which integrates the finite state machine and event-condition-action frameworks. This framework provides three products for system design: the SECA model–an abstract model of rules; the SECA state chart–a graphical visualization of rules; and the SECA rule description–a machine-readable format for practical deployment. The SECA framework offers several advantages, including convenient deployment, high efficiency, better logicality, and scalability. Simulation results demonstrate that the SECA framework enables autonomous decision-making in air confrontation scenarios and outperforms the event-condition-action framework in terms of computational time and cost-effectiveness. Furthermore, the generalization test in robot navigation tasks verifies its potential applicability to other domains with different background knowledge.