Assessing Cognitive Workload of In-Vehicle Voice Control Systems

Abstract

In-Vehicle Information Systems (IVIS) are becoming more accessible to drivers and contain more complex communication features. Voice control systems (VCS's) promise to be less distracting than visual-manual interfaces, but they still impose cognitive workload as drivers divert attention to using the voice interface instead of being attentive about safe driving. The goal of this dissertation is determine if the cognitive distraction induced by VCS's can be reliably measured using the Tactile Detection Response (TDRT) protocol. A contextual interview and two driving simulator studies were conducted. Findings from driving simulator studies showed that the TDRT was sensitive to changes in cognitive workload for VCS subtasks such as listening, speaking, visual search, confirmation, and waiting. The TDRT was less reliable in inferring the cogntive workload of VCS interactions containing errors (e.g. recognition error and system timeout). VCS's are multi-faceted interactions that require various cognitive processes complete. Understanding the cogntive workload of each cognitive process provides better guidance on how to improve the designs of VCS interactions.