Real-time fatigue monitoring system for motorcycle riders using sensors and a random forest algorithm.

Fatigue is a major determinant of motorcycle accidents, posing a critical threat to road safety by impairing riders' psychophysiological performance and increasing the crash risk. This study introduces a real-time fatigue monitoring system specifically designed for motorcycle riders, integrating physiological signal acquisition with machine learning classification. Heart rate variability (HRV) and galvanic skin response (GSR) sensors, established biomarkers of fatigue, were employed to continuously collect physiological data. The acquired signals underwent pre-processing to minimize noise and artefacts, followed by extraction of key features, including time-domain HRV indices and GSR conductance levels. These features were classified using a random forest algorithm, selected for robustness and accuracy in high-dimensional data contexts. The system discriminates fatigued from non-fatigued states in real time and delivers multimodal alerts to promote timely rest. Validation through laboratory and field trials demonstrated 84% accuracy, underscoring the potential of physiological monitoring with machine learning to enhance motorcycle rider safety.