Re-Envisioning On-Ground Aircraft Movement using Reinforcement Learning

Ground traffic congestion remains a significant challenge in the aviation industry, driven by stringent protocols, increased air traffic, and occasional controller errors, all of which contribute to delays in aircraft operations. Air Traffic Controllers (ATCOs) currently rely on Standardized Taxi Routes (STRs), which provide predefined taxi-in and taxi-out pathways. However, the limited flexibility of STRs, due to their finite number of published routes, constraints controllers' ability to optimize ground traffic flow. This study proposes a novel approach leveraging Reinforcement Learning (RL) to dynamically generate the most efficient taxi routes for aircraft in real-time. The proposed RL-based framework adapts to live ground conditions, enabling the system to adjust route clearances in response to situational changes. By reducing congestion and minimizing delays, this approach aims to alleviate ATCOs workload, enhance situational awareness, and improve overall operational efficiency during periods of high ground traffic density.