Driver Fatigue and Microsleep Detection via Eye Aspect Ratio (EAR): A Comparative Telemetry Analysis in High-Performance Simulation

This study evaluates the effectiveness of the Eye Aspect Ratio (EAR) algorithm for detecting driver fatigue and microsleep events within a high-fidelity driving simulator environment. The research aims to correlate ocular biotelemetry with operational performance and incident occurrence across varying driving dynamics and environmental conditions. Data was collected from 7 subjects (N=7) performing a 10.5 km technical route under two distinct states: alert and fatigued. To ensure algorithm robustness, a diverse range of vehicle platforms was utilized "Forza Horizon 5", including the BMW M2, Porsche 918 Spyder, Mini Cooper JCW, Honda Civic Type R, Ford Mustang, and Audi A1. The study further incorporates environmental and physiological variables, such as rain-induced visibility reduction and subjects operating under compromised health states (illness). Key findings indicate that alert states consistently maintained an EAR baseline between 0.35 and 0.38, characterized by rhythmic and deep blinking. In contrast, fatigued states showed a measurable baseline degradation (ptosis) to values below 0.29, signal compression, and erratic "blank stare" morphologies. Furthermore, critical events, including microsleeps and collisions, were preceded by sharp EAR drops, validating the algorithm’s predictive capacity for road safety. The results demonstrate that the EAR metric provides a reliable and non-invasive real-time indicator of driver vigilance, capable of detecting impairment caused by fatigue, illness, or environmental stress before a collision occurs.