Pilots' Task Identification Using Visual Attention Distribution

Pilots have different visual attention distributions when executing different tasks outlined in the Quick Reference Handbook (QRH). In this study, we use pilots' visual attention distribution to analyze their process of executing QRH tasks. We sampled the pilots' eye movement data by using eye-tracking devices and used a Markov Process model to extract features related to changes of visual attention distribution. A sliding window method was then used to describe the pilots' task switch process, and to identify the time point when pilots enter the procedure of a QRH task. We proposed the Adaptive Window Size (AWS) algorithm to overcome the limitations of the traditional sliding window method. The AWS algorithm dynamically adjusts the window size to fit the duration of each QRH task, enabling more accurate predictions of tasks. The results show that this model can accurately describe the whole process of pilots' QRH task execution and locate the time point when pilots enter the procedure of a QRH task. The meaning of this work is to provide a method to analyze the task execution process more precisely, and it is important for improving crew training, accident prevention, and flight safety.