Efficient Multiple Aircraft Conflict Resolution with Consensus Builder in Partially Observable Scenarios

The growth of the global economy has fueled the rapid expansion of the aviation industry. However, the increasing number of aircraft operating within limited airspace resources poses significant challenges to safe flight operations. In complex free-route airspace scenarios, resolving conflicts within unmanned aircraft clusters has become a critical technology to overcome airspace utilization bottlenecks. Particularly in the context of distributed control, aircraft must be capable of making effective maneuvering decisions based solely on locally observed information to enhance autonomy and cooperative efficiency.This paper addresses the conflict resolution problem for unmanned aircraft clusters under partially observable conditions by proposing an algorithmic framework that integrates a knowledge distillation model with multi-agent reinforcement learning. By introducing a consistency-aware cognitive auxiliary module, the approach enhances global situational awareness, optimizing decisionmaking efficiency and conflict resolution effectiveness. Simulation results demonstrate that the proposed algorithm significantly improves the conflict resolution capabilities of unmanned aircraft clusters in complex airspace environments, providing robust technical support for coordinated multi-aircraft operations.