EEG-based offline fatigue detection in Airbus A320 pilots during simulated night flights

Abstract

Fatigue is known to be a contributing human factor in civil aviation accidents. Although self-assessment of pilots evidently fails to evaluate whether a critical fatigue level is reached, little effort was made to unburden pilots by developing a suitable fatigue detection system (FDS). As electroencephalography (EEG) supply reliable and valid fatigue indicators and state-of-the-art EEG applications can provide both comfortability and online assessment, EEG is predestinated to be used for a FDS. The current study applied an approach from Yeo, Li, Shen and Wilder-Smith (2009) who successfully tested a FDS for car drivers. This EEG-based system made use of a machine-learning algorithm (support vector machines (SVM)) to achieve an impressive accuracy score of 99.3%. Here, simulated night flights with current Airbus A320 pilots were conducted to collect EEG data during autopilot and manual flight scenarios. Offline, the SVM algorithm was able to predict expert ratings with an accuracy of up to 92%. These results suggest that an EEG-based FDS is conceivable, although many challenges have to be met before its final application in real aviation contexts.