From Subjective to Objective: Developing EEG-Based Indicators for Competency Assessment in Low-Altitude Aviation Training

Abstract

The global advancement of the low-altitude economy has led to a significant increase in low-altitude aircraft operations and activities. Consequently, the selection and training of pilots for these aircraft have become focal points in both research and regulatory oversight. Traditional competency assessments often rely on subjective questionnaires, behavioral tests, or simulator training records. While these approaches can reflect macro-level performance, they exhibit limitations in objectivity, real-time capability, and the interpretation of underlying physiological mechanisms. Electroencephalography (EEG), a non-invasive physiological signal with high temporal resolution, offers a promising tool for pilot selection, training, and real-time monitoring of cognitive states. This study develops a resting-state EEG-based indicator to assess pilot competency. We designed a novel paradigm to collect multi-channel resting EEG from 20 participants. After preprocessing, signals were segmented into 4-second epochs. We extracted 14 features including power spectral density, differential entropy, and time-domain variance, analyzed across standard frequency bands. Using balanced samples per participant, we evaluated five classifiers via 10-fold cross-validation, comparing performance across bands, channels, and features. Results indicate that differential entropy features in the gamma band, particularly from the FP1 and P3 channels, demonstrated a significant ability to distinguish pilot competency. Classification accuracy based on single-channel gamma band DE features exceeded 80%. Model performance was further quantified using a confusion matrix and multiple metrics, such as accuracy, precision, recall, and F1-score. The study also discusses limitations related to sample size, model generalizability, and practical engineering applications, along with potential directions for improvement.