Until recently there was no generally accepted definition of distraction, and the most prominent discordance was that some researchers included cognitive inattention in the concept, while others reserved the term solely for visual distraction. In 2005 a large part of the First International Conference on Distracted Driving was dedicated to agreeing on a definition. The resulting definition clearly excludes long-term impairments like fatigue and alcohol intoxication, and states that the attention must be drawn away from driving towards something else in order to qualify as ”distraction”. A multitude of methods has been used to assess the prevalence and the types of driver distraction that occur, and to describe the consequences in terms of driving performance and crash involvement. There is strong agreement that distraction is detrimental for driving, and that the risk for crashes increases. Only recently the method of remote eye tracking has emerged, which enables real time identification of visual distraction. So far this method has mostly been used in driving simulators, and different algorithms that diagnose distracted drivers have been tested with promising results. Earlier research has shown that eye glances away from the road rarely exceed a duration of 2 sec. Most ”normal” glances range from about 0.7 sec. to slightly above 1 sec. In general, drivers rather opt for repeated glances instead of extending one single glance, if the secondary task demands attention for a longer period of time. It has been shown, however, that repeated glances have more detrimental effects on driving performance than a single glance of the same duration as one of the repeated glances. Apparently the drivers look away from the forward roadway again before they are completely back “in the loop”. Consequently, most algorithms that diagnose driver distraction based on glance behaviour do not only consider the most recent glance, but take the recent glance history into account. Some distraction mitigation strategies have been tested in driving simulators. The drivers were either advised to look back at the road, or the interaction with the secondary task was terminated by the system. The results of those studies were mixed, and it could not clearly be shown that the countermeasures tested improved driving performance. It has to be noted, however, that the results stem from driving simulator experiments, during which distraction was induced artificially. It is recommended to test both the algorithms used to diagnose driver distraction and the countermeasures in the field with naturalistic distraction. Generally it is important to focus research on naturalistic distraction because it is not clear how much artificially induced distraction makes the driver ”forget about” driving, or whether it rather resembles dual task performance in which the driver tries to maximise performance in both tasks and is well aware of the additional demands. A field study with real time eye tracking would be able to shed light on this question and also allow to evaluate a possible distraction countermeasure over a prolonged period of time. (Author/publisher)
Samenvatting