Objective Metrics for Assessing Visual Complexity of Vehicle Dashboards: A Machine-Learning Based Study

Dashboard is a central component of an in-vehicle information system (IVIS), and plays a crucial role in providing drivers with key information related but not limited to driving. With the expansion of the IVIS features, modern dashboards additionally integrate various new elements, which often leads to an increase in their visual complexity. Since high visual complexity of the dashboards threatens driving safety and performance, it is essential for researchers and designers to understand what objective features of the dashboards are related to their perceived visual complexity (PVC) so as to establish more cognitively efficient dashboards. In the present study, we refined the objective metrics of assessing visual complexity proposed in previous research and added two new dimensions, colors and animation, to better characterize recent development in the dashboard displays. We then utilized the indicators in the metrics to predict the dashboard PVC. Machine learning was innovatively applied, and the models were found to have stable performance. The study contributes reliable metrics and novel methodology to evaluate the visual complexity of the dashboards for the reference of future studies.