Aircraft taxiing trajectory prediction helps identify potential conflicts and collision risks in advance, ensuring safe and efficient airport operations. To highlight the important role of control system taxiing path instructions during aircraft taxiing, and to better integrate the influence of taxiing path instruction features on taxiing trajectories, this paper combines airport vector maps with taxiing path instructions from the control system and uses a depth-first search algorithm to capture the spatial features of taxiing path that aircraft focus on at the current moment, then uses Transformer encoder to capture the overall features of this sequence. To reflect the constraints of taxiing path instructions on future aircraft taxiing, we propose a Trajectory and Path Feature Modulator that uses the spatial features of taxiing path to modulate the aircraft’s motion state. Additionally, a weighted loss function is introduced, assigning different weights to different time points and appropriately amplifying the weight of the final moment to optimize the final displacement error. Verification using real aircraft taxiing trajectories from a major hub airport in China shows that the average displacement error for predicting position trajectories 8 s in the future is 2.405 m, the final displacement error is 3.112 m, and the average processing time is 13.04 ms, allowing for accurate and rapid prediction of taxiing trajectories, which contributes to ensuring safe operations in the flight area.
Aircraft Surface Taxiing Trajectory Prediction Fused with Path Instructions
Communic.Comp.Inf.Science
International Conference on Intelligent Computing ; 2025 ; Ningbo, China July 26, 2025 - July 29, 2025
2025-07-25
12 pages
Article/Chapter (Book)
Electronic Resource
English
GRU , Feature Modulation , Trajectory Prediction , Transformer Computing Milieux , Artificial Intelligence , Software Engineering/Programming and Operating Systems , Computer Science , Data Mining and Knowledge Discovery , Computer Communication Networks , Computer Imaging, Vision, Pattern Recognition and Graphics
Aircraft taxiing scheduling method based on taxiing conflict risk prediction
European Patent Office | 2022
|