Akademik Veri Yönetim Sistemi
COGNITION TECHNOLOGY & WORK, 2025 (SCI-Expanded)
This study proposes a monitoring and decision system for the real-time detection of operator fatigue in socio-technical work environments where system interaction and human performance are closely associated. The developed framework provides continuous monitoring of fatigue status through an effective real-time approach that supports shared autonomy and flexible decision support mechanisms. The measurement system, which features a user-centered ergonomic design, is equipped with sensors to measure skin conductance, movement acceleration and postural stability. A special simulation environment was created with attention-demanding tasks to simulate real-world operational conditions. Data collected from twenty-four participants were classified according to task performance and self-reported fatigue ratings. Physiological data were first filtered to remove noise and then analyzed using a deep learning model, achieving an F-measure of 85.3%. When the contribution of each sensor to the system performance was examined separately, the sitting balance obtained with the pressure-sensitive chair and the deterioration in micromovements emerged as reliable predictors of fatigue. The proposed system provides a low-cost, non-invasive and effective solution to detect operator fatigue without the need for a camera or wearable device. By combining behavioral modeling, ergonomic sensor data and machine learning, it supports human-AI collaboration in areas such as industrial inspection, transportation and defense, and contributes to the development of adaptable and resilient socio-technical systems.