A Tunnel Gaussian Process Model for Learning Interpretable Flight's Landing Parameters
This work aims to improve flight safety by providing interpretable insights into flight dynamics for aircrew and air traffic controllers, which is an incremental improvement to existing risk reduction technologies.
The paper proposes a data-driven method, the Tunnel Gaussian Process (TGP) model, to learn and interpret flight approach and landing parameters from A-SMGCS data. TGP demonstrates superior modeling performance compared to existing methods on synthesized trajectory datasets and provides interpretable probabilistic descriptions of landing dynamics when applied to operational data.
Approach and landing accidents have resulted in a significant number of hull losses worldwide. Technologies (e.g., instrument landing system) and procedures (e.g., stabilized approach criteria) have been developed to reduce the risks. In this paper, we propose a data-driven method to learn and interpret flight's approach and landing parameters to facilitate comprehensible and actionable insights into flight dynamics. Specifically, we develop two variants of tunnel Gaussian process (TGP) models to elucidate aircraft's approach and landing dynamics using advanced surface movement guidance and control system (A-SMGCS) data, which then indicates the stability of flight. TGP hybridizes the strengths of sparse variational Gaussian process and polar Gaussian process to learn from a large amount of data in cylindrical coordinates. We examine TGP qualitatively and quantitatively by synthesizing three complex trajectory datasets and compared TGP against existing methods on trajectory learning. Empirically, TGP demonstrates superior modeling performance. When applied to operational A-SMGCS data, TGP provides the generative probabilistic description of landing dynamics and interpretable tunnel views of approach and landing parameters. These probabilistic tunnel models can facilitate the analysis of procedure adherence and augment existing aircrew and air traffic controllers' displays during the approach and landing procedures, enabling necessary corrective actions.