Driver monitoring system as Road Focus Reminder

Human errors due to distraction are a direct cause or a contributing factor in a majority of vehicle crashes and impacts. Visual-manual secondary tasks are critical according to naturalistic driving studies. NHTSA’s recently proposed Design Guidelines with purpose to limit potential driver distraction from secondary, non-drivingrelated, visual-manual tasks performed on integrated electronic devices. With external cameras and radars, some of the crashes caused by distraction can be mitigated or even avoided thanks to active safety functions. An attention sensor could increase the effectiveness of such systems and generate positive side effects. Moreover, such a sensor will be useful in future cars with partially or fully automated driving. A prototype system for real-time measurement of driver visual attention has been developed by Autoliv. The sensor was installed in 10 vehicles by ÅF-Technology (ÅF) and Volvo Car Corporation (Volvo Cars), and used in everyday traffic. Drivers/users were Volvo Cars employees with a company car leasing contract. Cars were also equipped automatic emergency braking; adaptive cruise control and lane keep assist. The overall aim of the project was to evaluate the possibility to influence the driver’s visual behavior toward a reduction of critically long glances away from the road. For this purpose a naturalistic driving study / Field operational test was carried out with an online warning alerting the driver if s/he was looking away from the roadway for too long. The experimental protocols used was within group ABA design, meaning data was collected in three phases, baseline, treatment and repeated baseline. This allows for testing the effect of warnings compared to no warnings, and a possibility to see if the effects of warnings persist after removal. In addition to analysis of distraction warnings, uniquely broad naturalistic data about driver glance behavior has been collected. Data show great similarities to previous studies where video clips of driver collected in naturalistic settings have been manually annotated for eyes-off-road, indicating a basic level of validity. It was shown that driver off road glance duration is dependent on among other things, driving speed, daylight, use of Adaptive Cruise Control (ACC), lead vehicle presence and weekend or weekday driving. There was also a tendency of reduced number of warnings when the distraction warning was active, compared to baseline performance.