Aircraft landing is a complex and challenging flight phase in which pilots are required to allocate attention efficiently to the surrounding environment. A comprehensive understanding of pilot situation awareness (SA) is needed for successful landing. This study was to predict pilot SA during landing using eye tracking data. The experiments were carried out with 5 repetitive simulated landings for four expert and four novice pilots using eye tracking equipment. Three eye tracking features (visit frequency, dwell time ratio, scan path entropy) were developed for reflecting three level of SA model (perception, comprehension, projection). Prediction of SA was performed by developing multiple regression model. Visit frequency of expert pilots was 138%, 47%, 85%, 67%, 117%, and 91% higher than novice pilots in RPM, VVI, altimeter, heading, airspeed, and attitude areas of interest (AOIs) respectively; while 50% and 33% lower in runway and outside AOIs respectively. Dwell time ratio of expert pilots was 38% and 42% higher than novice pilots in runway and outside AOIs respectively; while 62%, 62%, and 65% lower in altimeter, airspeed, and attitude AOIs respectively. Scan path entropy of expert pilots was 33% higher than novice pilots in outside AOI; while 29% lower in attitude AOI. Coefficient of determination for the prediction model for SA was 80.6%. The results of this study can be used as objective data of strategy establishment or training feedback for novice pilots.

• 김성호(공군사관학교 시스템공학과, Department of Systems Engineering, Air Force Academy) | Sungho Kim