LLM4Delay: Flight Delay Prediction via Cross-Modality Adaptation of Large Language Models and Aircraft Trajectory Representation
arXiv:2510.23636v4 Announce Type: replace-cross Abstract: Flight delay prediction has become a key focus in air traffic management (ATM), as delays reflect inefficiencies in the system. This paper proposes LLM4Delay, a large language model (LLM)-based framework for predicting flight delays from the perspective of air traffic controllers monitoring aircraft after they enter the terminal maneuvering area (TMA). LLM4Delay is designed to integrate textual aeronautical information, including flight data, weather reports, and aerodrome notices, together with multiple trajectories that model airspace
The increasing sophistication of large language models and the growing need for optimized air traffic management are converging to enable advanced predictive solutions.
This development indicates a tangible application of AI, specifically LLMs, to critical infrastructure management, moving beyond general text generation to tangible operational efficiency and safety improvements.
LLMs are being adapted and integrated with complex, multi-modal data for real-world, high-stakes prediction tasks, demonstrating their utility in operational domains like air traffic control.
- · Air traffic management systems providers
- · Airlines
- · Airports
- · AI/ML model developers
- · Traditional air traffic control methodologies
- · Manual data analysis processes
Flights become more efficient and punctual, reducing operational costs and passenger frustration.
Increased reliance on AI in safety-critical systems prompts new certification and regulatory frameworks for AI-driven operations.
The success of LLM4Delay could catalyze similar cross-modality AI adaptations in other complex transportation and logistics networks, like maritime or rail, leading to a new era of predictive infrastructure management.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.AI