A reinforcement learning-based routing approach for optimizing airport ground movement

作者：Wei Yang, Hong Liu, Aoli Jia, Long‐Qing Chen, Hong‐Yu Yang · 发表于：Transportmetrica B Transport Dynamics · 年份：2025 · DOI：10.1080/21680566.2025.2496824 · 被引用次数：3 · 研究领域：Air Traffic Management and Optimization、Traffic control and management、Autonomous Vehicle Technology and Safety

As global air passenger traffic increases, optimizing airport ground operations becomes crucial. Recent studies have largely focused on combinatorial optimization and sequential planning methods. However, the former's lengthy computation times hinder real-time decision-making, while the latter's sequential path planning limits flexibility and optimality. This paper introduces an autonomous approach for real-time optimization of aircraft taxiing routes, circumventing the restrictions of sequential planning. Our model utilizes real-time data on aircraft motion and taxiway layouts, accounting for operational constraints and dynamic scenarios. By adapting the Proximal Policy Optimization (PPO) algorithm to a distributed, multi-agent framework, we enhanced network architecture experience collection processes, and agent coordination. Validated with operational data from Tianfu International Airport, our algorithm demonstrated improved robustness, adaptability, and execution capabilities over traditional sequential methods. These results highlight the potential for integrating this approach into automated airport ground traffic management systems, promising significant gains in efficiency and flexibility.