A Hybrid Approach Based on Behavioural and Physiological Data for Driver Monitoring Systems

Driver Monitoring Systems are gaining increasing interest from car-makers, being featured by technologies capable to improve road safety levels and prevent accidents. Currently, these systems are mostly passive, based on distance traveled and travel time. The market trend is moving towards active systems, featured by popular approaches, such as behavioral and physiological. This article proposes a Driver Monitoring System to detect driver's drowsiness with a hybrid technique that uses both approaches. The first performs a driver's face analysis by monitoring ocular variables; the second is based on biometric signals, such as Heart Variability and Galvanic Skin Response. A distributed hardware/software architecture has been designed in order to develop real-time data acquisition and a fuzzy-based inference. Comparison tests between the hybrid approach and the two single ones showed the benefits of the proposed method in estimating driver's drowsiness, considering a subjective estimation conducted during driving sessions with the CARLA simulator.