Abstract

Bio-signal recordings that reflect the operating conditions of the physiological systems provide several useful metrics to determine the dynamics of internal states in human body. In this context, this study is aimed to identify the useful metrics indicating the real-time stress level of drivers. Totally 10 of drivers’ bio-signal datasets obtained from MIT-BIH PhysioNet Multi-parameter Database were used to quantify the stress level of drivers. Each dataset includes several bio-signal recordings gathered from drivers during the drives conducted in a set path where 3 different levels of stress likely occur through over 20 miles of open roads. The recordings have markings indicating the time intervals between the driving segments corresponding to a number of city and highway driving periods with initial and final rest periods. Evaluations have been completed by using the available segment based arrays of instantaneous heart rate (IHR), hand based skin conductance (HSC), foot based skin conductance (FSC) and electromyography (EMG) signals. For further evaluations, the segment based data arrays including instantaneous respiratory rate (IRR) and average number of contractions/minutes (CPM) have been derived from the respiratory and EMG signals, respectively by using a peak detection algorithm. Statistical comparisons using the overall mean and mean values of the IHR, HSC, FSC, EMG, IRR and CPM data arrays showed that with an optional exception of IRR all of these metrics can be identified as useful parameters in the future car technology to determine the dynamic stress level of drivers.

Key words: Driving stress of drivers, dynamic stress level, electromyography, heart rate, respiration rate, skin conductance.