A Method for Detection and Evaluation of Driver Distraction Induced by In-Vehicle Information Systems

Steep improvement of an in-vehicle info- and entertainment systems has a positive impact on vehicle control, comfort, safety, etc. Nevertheless, it also leads to a multitasking load increase on the in-vehicle information system due to fundamental problems of driver distraction. In this paper, a method for detection and evaluation of driver distraction induced by the driver's secondary activity is developed. The methodology is based on the machine learning and computation intelligence algorithms blend, which includes a driver model, a driver distraction detector, and a fuzzy logic evaluator. Several data fitting algorithms efficient for nonlinear regression are designed and are compared on the accuracy of the driver performance prediction. The method is verified by the driver-in-the-loop experiment with thirty participants on an advanced vehicle simulator. Driver's interaction with the commercial in-vehicle information system is exploited as a secondary distractive task.