Naturalness framework for driver-car interaction

Abstract

Automobile dashboards are evolving into intelligent largely screen-based computer interfaces. Recent evidence suggests unnatural aspects of some secondary systems (including infotainment systems). Naturalness of interaction is a minority discipline not yet applied to the automobile; while automotive interface research is a mainly quantitative discipline that has not yet applied a naturalness approach. To advance the field, a measurement scale based on rigorous qualitative exploration of natural-feeling interaction with secondary controls was required. Study 1 used ethnographic interview with Contextual Inquiry inside 12 ordinary drivers’ cars, to investigate natural-feeling aspects of past, present and future driver-car interactions. Thematic analysis suggested a framework of ten characteristics. Half concerned control and physicality; half concerned perceived socio-intelligent behaviours of the car. Study 2 involved intensive exploratory workshops with ten drivers comprising Think Aloud, artefact modelling and focus groups, to explore natural-feeling interaction with secondary controls in different ways. The resulting thematic framework comprised 11 characteristics in four categories: familiarity/control, physical connection, low visual/cognitive demand, and humanlike intelligence and communication. Study 3 comprised two ethnographic participant observations. Eight drivers were observed interacting with their controls during long road journeys. Twenty-two drivers were observed interacting verbally with futuristic ‘intelligent’ secondary systems while driving on public roads. Design guidelines relating to physicality, usability, automation, and humanlike communication were formulated. Study 4 converted all the qualitative findings into a questionnaire comprising 46 bipolar five-point scales. Eighty-one drivers used it to rate one control in their cars. Correlation and factor analyses revealed three underlying factors and 14 items suitable for the first industrially applicable measurement scale for driver-car naturalness. These items concern perceived helpfulness, politeness, competence, predictability, control, ease, mental demands, intuitiveness, ‘realness’, instantaneousness, communication, logical location, mapping and 'affordance'. Initial testing found acceptable validity. The conclusion recommends further data collection, expanded validity testing, and potential applications to self-driving cars.