A model of driver’s visual distraction

Nowadays, one of the most challenging issues faced by Advanced Driving Assistance Systems (ADAS) is how to “adapt” to the current context in order to safely support the driver, i.e., how to properly react to changes in the internal and external conditions. On this respect, the driver’s psychological state is considered a relevant context aspect to be taken into account. In this study, we present a model of the driver’s visual distraction that can be exploited for designing the adaptation behavior of an ADAS. The model has been developed by leveraging ecological data collected through an in-vehicle experimental campaign. Thirty participants have volunteered in the experiment. Participants have been driving on a test site including a highway part and a extra-urban part. The distraction has been induced by an interfering task consisting in a visual search performed on a secondary touch screen (Surrogate Reference Task (SURT)). The distraction task resembles the interaction of the driver with a real touch screen device on the car dashboard. By means of a proper vehicle equipment configuration (internal camera, vehicle CAN bus and secondary task), we have been able to collect data about the driving behavior (like the steering angle, vehicle speed, position of the vehicle in the lane, etc.), the driver’s head orientation, and the execution of the secondary task (e.g., response times, correct answers, etc.). We have analyzed the data in order to compare the driving behavior with and without the SURT in the different road scenarios. The analysis results show that the distraction task significantly affects the driving performance in both scenarios. We have exploited such results to derive the model of the driver’s visual distraction based on vehicle dynamics.