Abstract:

Despite recent gains in the accuracy and flexibility of voice interfaces, speech-enabled in-vehicle information systems (IVIS) still impose a significantly higher cognitive load than vehicle operation alone [6.7]. This results in degraded driving performance while carrying out common information-retrieval (IR) tasks such as finding a particular point of interest (POI) from a navigation database or a particular song from a music library. This paper proposes a reorientation of the IVIS interface around domain-scoped searches and contextual commands rather than around hierarchical menus and global commands. We believe that this design will reduce IR task time while also reducing cognitive load, thereby encouraging safer driving.