Sleepiness is a common human factor among truck drivers resulting from sleep loss or time of day and causing impairment in vigilance, attention, and driving performance. While driver sleepiness may be associated with increased risk on the road, sleepy drivers may drive more cautiously as a result of risk-compensating behaviour. This endogeneity has been overlooked in the previous driver behaviour studies and may provide new insight into the effects of sleepiness on driving performance. In addition, the Karolinska Sleepiness Scale (KSS) has been widely used to quantify sleepiness. However, the KSS is a subjective self-reported measure and is reliant on honest reporting and understanding of the scale. An alternative way of quantifying sleepiness is using drivers’ heart rate and correlating it with their sleepiness. While recent advances in data collection technologies have made it possible to collect heart rate data in real-time and in an unobtrusive way, their application in measuring sleepiness particularly among truck drivers has been unexplored.

This study aims to address these gaps and contribute to analytic methods in road safety research by collecting truck drivers’ heart rate data in real-time, measuring sleepiness from those data, and using it in an instrumental variable modelling framework to investigate its effect on driving performance. To this end, a driving simulator experiment was conducted in Belgium and heart rate data were collected for 35 truck drivers via sensors installed on the steering wheel of the simulator. Additional demographic data were collected using a questionnaire before the experiment. An instrumental variable model consisting of a discrete binary logit and a continuous generalized linear model with grouped random parameters and heterogeneity in their means was then developed to study the effects of driver sleepiness on headway. Results indicate that age, years of holding driver licence, road type, type of truck transport, and weekly distance travelled are significantly associated with sleepiness among the participants of this study. Sleepy driving is associated with reduced headway for 30.5% of the drivers and increased headway for the other 69.5%, and night-time shift is associated with such varied effects. These findings indicate that there may be group- or context-specific risk patterns which cannot be explicitly addressed by hours of service regulations and therefore, transport operators, driver trainers and fleet managers should identify and handle such context-specific high risk patterns in order to ensure safe operations.

Driver sleepiness and fatigue are important contributors to many transport incidents and significantly increase crash risk. Recently, detection systems have been developed which aim to monitor the state of the driver and detect increasing levels of fatigue. However, there has been less focus on appropriate intervention strategies for drivers once fatigue and sleepiness have been detected. This paper describes the i-DREAMS fatigue intervention strategies, which aim to keep drivers within a safe driving zone. Interventions will be provided both in real-time and post-trip and can be customized to be used with a variety of transport modes. Real-time interventions will measure fatigue through trip duration, and driver sleepiness through heart-rate variability (HRV) information, obtained by means of sensors in the steering wheel or a wearable device, and attributed to Karolinska Sleepiness Score (KSS) bands. Thresholds for warnings will map onto phases of a ‘Safety Tolerance Zone’ and will be dynamic – changing as the driver state and driving situation develops. Post-trip interventions will aggregate data throughout the duration of the drive and aim to provide customized feedback and coping tips related to driver levels of fatigue and sleepiness, to improve driving behavior. Goals and challenges will add a gamified aspect to the post-trip interventions. The next stage of the development of the i-DREAMS fatigue intervention strategy is to test the concept in a series of simulator and field trials. Future research should explore acceptance and compliance of interventions and frequency of alerts.