Design implications for visual-manual interfaces in cars

Abstract: During the past decades there has been an extensive evolution in computer and communication technologies. In the last ten years these technologies have become an integrated part of our cars via in-vehicle information systems. Considerable efforts are put into research and development of the interface between the in-vehicle information systems and the driver. Due to the fact that vision is imperative in driving, the focus is in particular on the visual-manual interface and on apt usability evaluation methods. The purpose of this thesis is to study design implications for the development of visual-manual interfaces in cars. A point of departure is that the practice of the usability concept, usability evaluation methods, and compliance tests has implications for the design process. Four papers are appended to the thesis. Paper I focuses on differences between rotary control and touch screen interfaces in terms of driving simulator and glance metrics. Paper II investigates correlations between data from customer evaluations and expert evaluations for visual-manual interfaces. Paper III searches for the presence of and possible explanations for individual glance strategies in a driving simulator study. Paper IV investigates effects of individual glance strategies on the NHTSA visual-manual distraction test in a driving simulator set-up. In the thesis it is concluded that the optimal input device is dependent on the task being performed, that is, the touch screen is more efficient for alphanumeric input while rotary control performs list scrolling more efficiently. In addition, drivers seem to adapt their glance behaviour based on the display position. It is evident that efficiency metrics correlate between several usability evaluation methods. Hence, it is possible to generalise results from expert evaluations at an early design phase, to customer survey samples based on car ownership. Concerning glance behaviour, it is clear that individual glance strategies exist. Especially interesting are two glance strategies,long glancers and frequent glancers, that affect the results of the NHTSA compliance testing. These glance strategies have to be controlled for in the test situation. Finally, a framework is proposed that combines a classification of visual-manual actions with different visual fixations. The task analysis framework has the potential to analyse actions in terms of required visual resources, thus already providing an evaluation of visual-manual interfaces in the concept phase.