Tag
This paper introduces CaTR, a value-decomposed reinforcement learning framework for real-time multi-aircraft taxiway routing that uses hierarchical foresight traffic representation to balance safety and efficiency.