An Evaluation of the IVIS-DEMAnD Driver Attention Demand Model

This paper presents results of a study conducted to apply and evaluate the In-Vehicle Information System (IVIS) DEMAnD Model developed recently by the Virginia Polytechnic University's Center for Transportation Research for the Federal Highway Administration. This software-based model allows vehicle design engineers to predict the effects an in-vehicle information system might have on driver performance. The model was exercised under nine different driver attention task levels ranging from simple, such as glancing into a side view mirror, to complex, such as operating an in-vehicle navigation system. The nine driver tasks were evaluated using three different vehicle configurations and two levels of driver-roadway complexity. In addition, real-world information on driver visual performance was also collected during four different tasks for comparison with model predictions of these same functions. The comparison of model prediction for maximum number of glances, total task performance time, and a model-rating feature called figure of demand for each of the tasks indicated:

1. 1
   Driver task performance behavior is influenced substantially more by differences in the level of driver/roadway/traffic combinations, than by differences in test vehicle configurations.
2. 2
   The driver performance data collected under actual driving conditions compared very well with the model predictions.
3. 3
   Overall, the IVIS DEMAnD model appears to be a good early attempt at modeling the effect on driver performance of in-vehicle IT (Information Technology) systems in general.