Real-Time Detector of Human Fatigue

The goal is to develop and validate an automated, real-time, fatigue detection system for operators seated at workstations. The system can be used for alerting the operator, self-assessment, supervision, and scheduling. The operational environments of interest include stationary and airborne command centers. The proposed system will combine (1) a well-validated mathematical model of circadian rhythm and performance, (2) wrist mounted actigraph and lux sensors used to initialize the circadian rhythm model, (3) a non-intrusive (and non-invasive to the primary task) visual face and eye monitoring system that will measure eye point angle, blink rate, percentage eye closure, and other physiological metrics, and (4) signal processing methods such as the extended Kalman filter that will optimally combine the model's fatigue prediction with the real time physiological measurements. The hypothesis is that this combined system will result in improved detection of operator fatigue. A Phase I experimental pilot study will demonstrate the system and test the hypothesis using a small subject population. A set of cognitive and psychomotor tasks will be used to provide an objective performance measure. A larger experimental study will be conducted in Phase II using multiple consoles in a simulated operation center.