Answers are grounded in this paper and cite their sources.
This paper presents an adaptive data-driven approach to modeling air traffic behavior to support the efficient monitoring and safety of airport terminal area operations. Addressing the limitations of static modeling in evolving transportation systems, the proposed framework continuously learns trajectory positional deviations from real-world data. It dynamically updates to incorporate new flight data as it becomes available. This incremental learning capability supports sustained relevance and flexibility without the computational burden of accessing or reprocessing raw historical flight databases. The methodology relies on a machine-learning algorithm that captures trajectory deviations and identifies relevant patterns using a modified Fréchet framework. Crucially for safety-critical applications, the model ensures explainability by grounding clusters in intelligible flight-path metrics, such as physical distances and shapes, rather than abstract representations. Well-established objective clustering and modeling metrics guide the optimization process. Experimental results across multiple airports demonstrate robust performance equivalent to or better than that of benchmark approaches. These findings confirm the framework’s value for Air Traffic Management (ATM) applications, demonstrating its use in airspace optimization and its potential applicability to broader domains, including maritime and vehicle trajectory modeling.
No takes yet. Share an insight, caveat, or question.
Chiaratti et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: