Multi-modal assessment of on-road demand of voice and manual phone calling and voice navigation entry across two embedded vehicle systems

One purpose of integrating voice interfaces into embedded vehicle systems is to reduce drivers' visual and manual distractions with 'infotainment' technologies. However, there is scant research on actual benefits in production vehicles or how different interface designs affect attentional demands. Driving performance, visual engagement, and indices of workload (heart rate, skin conductance, subjective ratings) were assessed in 80 drivers randomly assigned to drive a 2013 Chevrolet Equinox or Volvo XC60. The Chevrolet MyLink system allowed completing tasks with one voice command, while the Volvo Sensus required multiple commands to navigate the menu structure. When calling a phone contact, both voice systems reduced visual demand relative to the visual-manual interfaces, with reductions for drivers in the Equinox being greater. The Equinox 'one-shot' voice command showed advantages during contact calling but had significantly higher error rates than Sensus during destination address entry. For both secondary tasks, neither voice interface entirely eliminated visual demand. Practitioner Summary: The findings reinforce the observation that most, if not all, automotive auditory-vocal interfaces are multi-modal interfaces in which the full range of potential demands (auditory, vocal, visual, manipulative, cognitive, tactile, etc.) need to be considered in developing optimal implementations and evaluating drivers' interaction with the systems. Social Media: In-vehicle voice-interfaces can reduce visual demand but do not eliminate it and all types of demand need to be taken into account in a comprehensive evaluation.