Multimodal Data-Driven Suitable Pilot Selection
Predicting the success of promising pilot candidates throughout the hiring process is a substantial part of a company’s success. However, oftentimes this process is very time-consuming and only restricted information is available. For pilot selection, the standard assessments are typically based on flight instructors performance ratings. Nevertheless the interrater reliability might vary substantially and the predictive power of this procedure is rather medium.
Therefore, this project taps into the broad array of psychophysiological measuring systems to enhance traditional performance data that is collected within a flight simulator environment and to pin down the most promising predictors of pilot success. Trainees complete thoroughly designed missions which aim to assess different psychological and cognitive attributes as well as flight maneuvers which have been identified as particularly relevant for their future success as pilots. Data collected during this initial assessment is then used to model a predictive index that is later evaluated based on real-life flying performance and pilot success.
Insights and Outcomes
The multimodal flight-behavioral and physiological based flight simulator approach yielded an accuracy of 87% in predicting the success of a future jet pilot. This accuracy is significantly higher than purely cognitive test batteries or flight instructor based performance ratings.