Evaluating distraction of in-vehicle information systems while driving by predicting total eyes-off-road times with keystroke level modeling

Abstract

Increasingly complex in-vehicle information systems (IVIS) have become available in the automotive vehicle interior. To ensure usability and safety of use while driving, the distraction potential of system-associated tasks is most often analyzed during the development process, either by employing empirical or analytical methods, with both families of methods offering certain advantages and disadvantages.The present paper introduces a method that combines the predictive precision of empirical methods with the economic advantages of analytical methods. Keystroke level modeling (KLM) was extended to a task-dependent modeling procedure for total eyes-off-road times (TEORT) resulting from system use while driving and demonstrated by conducting two subsequent simulator studies.The first study involved the operation of an IVIS by N = 18 participants. The results suggest a good model fit (R2Adj. = 0.67) for predicting the TEORT, relying on regressors from KLM and participant age. Using the parameter estimates from study 1, the predictive validity of the model was successfully tested during a second study with N = 14 participants using a version of the IVIS prototype with a revised design and task structure (rPred.-Obs. = 0.58). Possible applications and shortcomings of the approach are discussed.