Evaluating driver attention and driving behaviour: Comparing controlled driving and simulated driving

Emerging in-vehicle systems have turned the contemporary car into a human-computer interaction context that has its own set of rules and challenges. Interacting with in-vehicle systems while driving a car can greatly affect the driving performance and have been shown to be the cause of many road accidents. Evaluation of in-vehicle systems is a subject of much interest to developers and researchers. One of the major issues is how to evaluate; is there added value in taking your evaluation on the road or is simulated driving sufficient? This paper examines differences and similarities between taking in-vehicle systems to a track or to the laboratory by investigating the results (and costs associated) from two driving settings for in-vehicle systems evaluation; one on a test track and one using a lightweight driving simulator. Our results show that the two settings do seem to lead to a number of similar results. However, our results indicated that controlled driving yield more frequent and longer eye glances compared to simulated driving and driving errors were more common in simulated driving.