A Novel Transformer-based Trajectory Options Generator for Collaborative Air Traffic Flow Management

Demand for air travel has increased significantly in recent decades, possibly exceeding capacity. Researchers have made efforts to achieve demand and capacity balance (DCB), such as Ground Delay Programs (GDPs) and Collaborative Trajectory Options Program (CTOP). However, previous trajectory design works in Collaborative Air Traffic Flow Management (C-ATFM) provide optimal trajectory based on abstract airspace structures built by human experience, and it may be difficult to accurately capture the complex flight operation environment. Therefore, this paper proposes a Transformer-based trajectory options generator (TTOG) for collaborative air traffic flow management to provide AUs with effective trajectory options, thus improve the performance of ATFM. Specifically, the generator utilizes the Transformer architecture to learn trajectory planning knowledge from historical data and generate a series of candidate trajectories that conform to the principles of trajectory planning. Then we formulate the 4D trajectory planning problem as an integer linear programming model to obtain the global 4D trajectories and minimize the total delay costs. Experiment results suggest that the generator could provide effective trajectory options and significantly reduce total flight delay costs.