Explainable RL for Air Traffic Control

A new arXiv preprint (2607.22525) explores explainable reinforcement learning (RL) for air traffic control (ATC). The study trains an RL agent to reroute flights around no-fly zones in a simplified ATC environment, aiming to build trust through explainability. The work addresses the challenge of integrating AI into high-stakes domains like aviation, where transparency is critical for human-AI collaboration. The agent's decisions are analyzed using explainability techniques to clarify its reasoning. This research contributes to the broader effort of making deep learning systems more interpretable in safety-critical applications.