Learning to Land Anywhere: Transferable Generative Models for Aircraft Trajectories
Olav Finne Praesteng Larsen, Massimiliano Ruocco, Michail Spitieris, Abdulmajid Murad, Martina Ragosta
TL;DR
This work tackles data scarcity in aviation trajectory generation by evaluating transfer learning from a data-rich airport to a data-scarce one. It systematically compares four generative model families—Diffusion Models, Flow Matching, and their Latent variants (LDM, LFM)—for cross-airport transfer from Zurich to Dublin. Diffusion-based approaches deliver the strongest transfer, achieving competitive results with only a small amount of Dublin data and reaching baseline performance at moderate data budgets; latent variants offer additional gains, while Flow Matching shows more limited improvements. The findings demonstrate that transferable generative knowledge can enable realistic synthetic trajectory generation in data-poor ATM environments, with implications for scalable simulations and what-if analyses.
Abstract
Access to trajectory data is a key requirement for developing and validating Air Traffic Management (ATM) solutions, yet many secondary and regional airports face severe data scarcity. This limits the applicability of machine learning methods and the ability to perform large-scale simulations or "what-if" analyses. In this paper, we investigate whether generative models trained on data-rich airports can be efficiently adapted to data-scarce airports using transfer learning. We adapt state-of-the-art diffusion- and flow-matching-based architectures to the aviation domain and evaluate their transferability between Zurich (source) and Dublin (target) landing trajectory datasets. Models are pretrained on Zurich and fine-tuned on Dublin with varying amounts of local data, ranging from 0% to 100%. Results show that diffusion-based models achieve competitive performance with as little as 5% of the Dublin data and reach baseline-level performance around 20%, consistently outperforming models trained from scratch across metrics and visual inspections. Latent flow matching and latent diffusion models also benefit from pretraining, though with more variable gains, while flow matching models show weaker generalization. Despite challenges in capturing rare trajectory patterns, these findings demonstrate the potential of transfer learning to substantially reduce data requirements for trajectory generation in ATM, enabling realistic synthetic data generation even in environments with limited historical records.