This paper describes the research conducted as part of the development of a driver interface for warning drivers about low oil pressure and worn tyres. In the first experiment, 27 drivers were asked questions that examined their knowledge of components that might fail. For about half of the items, their knowledge was inadequate. In a second experiment, 60 drivers selected the words they preferred for nine prototypical warnings from columns of choices. This new method allowed a large set of messages to be rapidly produced from a limited data set. Drivers preferred command messages (what to do) over error messages (what was wrong). In the third experiment, 20 drivers responded to 15 candidate messages. While most messages were well understood, this experiment identified a few problems with specific warnings and the vehicle mimic.
Abstract