Monitoring eye movements in real flight conditions for flight training purpose

Abstract

Eye-tracking is a relevant technique to investigate the pilot’s cognitive performance. It can enhance pilots’ training and be used foron-line monitoring for smart cockpits. Most of the studies are conducted in flight simulators which may limit the interpretation of theeye-tracking data. In this study, we investigate the possibility to measure eye movement in real flight. We conducted an experiment with 7 pilots performing two traffic patterns and basic flight maneuvers in a real light aircraft. We analyzed the distribution of attentionover the main areas of interest (flight instruments) during the different flight phases. These data were confronted with operationalprocedures and discussed with flight instructors in terms of flight safety and recommendations for training. Also, a classifier trainedon the first traffic pattern could predict the three phases (take-off, downwind, and landing) of the second one with a mean accuracy of70%