The evolution of fatigue in remote tower controllers: evidence from eye-tracking analysis

With the increasing deployment of remote tower operations (RTO) in air traffic management, understanding how fatigue evolves and manifests during task execution has become essential for ensuring operational safety. This study investigates the dynamic effects of active fatigue on eye movement behavior in remote tower controllers using a simulated RTO environment. 13 trainee controllers completed identical control tasks under alert and fatigued states, during which 8 eye movement features spanning saccade, fixation, blink, and pupil dimensions are continuously recorded. To capture temporal trends while accounting for inter-individual variability, generalized additive mixed model (GAMM) is employed to model feature evolution over task duration. The results revealed that fatigue induced systematic alterations in eye movement behavior. In the fatigued state, average saccade speed increased significantly and exhibited a clear linear upward trend with task duration, whereas fixation count and saccade count decreased and blink count increased, all showing pronounced nonlinear temporal fluctuations. Average blink duration was significantly higher under fatigue, while average fixation duration remained relatively stable across all states. By contrast, pupil diameter steadily declined in the fatigued state, showing a contrasting temporal trend compared with the alert state. These findings demonstrate that fatigue during RTO induces specific alterations in eye movement patterns with a degree of task dependency. Moreover, the results indicate that eye movement features are influenced by multiple factors, underscoring the need to integrate fatigue-sensitive features in order to improve the accuracy of fatigue detection.