Exploiting Blood Volume Pulse and Skin Conductance for Driver Drowsiness Detection

Attention loss caused by driver drowsiness is a major risk factor for car accidents. A large number of studies are conducted to reduce the risk of car crashes, especially to evaluate the driver behavior associated to drowsiness state. However, a minimally-invasive and comfortable system to quickly recognize the physiological state and alert the driver is still missing. This study describes an approach based on Machine Learning (ML) to detect driver drowsiness through an Internet of Things (IoT) enabled wrist-worn device, by analyzing Blood Volume Pulse (BVP) and Skin Conductance (SC) signals. Different ML algorithms are tested on signals collected from 9 subjects to classify the drowsiness status, considering different data segmentation options. Results show that using a different window length for data segmentation does not influence ML performance.