Performance and acceptance evaluation of a driver drowsiness detection system based on smart wearables

Abstract

Current systems for driver drowsiness detection often use driving-related parameters. Automated driving reduces the availability of these parameters. Techniques based on physiological signals seem to be a promising alternative. However, in a dynamic driving environment, only non- or minimal intrusive methods are accepted. In this work, a driver drowsiness detection system based on a smart wearable is proposed. A mobile application with an integrated machine learning classifier processes heart rate from a consumer-grade wearable. A simulator study (N=30) with two age groups (20-25, 65-70 years) was conducted to evaluate acceptance and performance of the system. Acceptance evaluation resulted in high acceptance in both age groups. Older participants showed higher attitudes and intentions towards using the system compared to younger participants. Overall detection accuracy of 82.72% was achieved. The proposed system offers new options for in-vehicle human-machine interfaces, especially for driver drowsiness detection in the lower levels of automated driving.