Prediction of Drivers’ Visual Strategy Using An Analytical Model

The proposed research addresses the problem of predicting driver's visual behavior. Notably, this issue is of great importance for road safety because it provides information about the true perception of the road environment. This need is particularly felt when the real situation is so complex that it cannot be represented by simple analytical audits contained in the road international standard. Therefore, with the help of an instrumented vehicle, we conducted some trials to identify relevant data useful to recognize the road environment as perceived by drivers. Based on this data, we prepared a linear model that, properly calibrated, would enable us to predict the drivers' visual strategy. The results were encouraging, since a model so simple in structure as well as in the number of variables involved, could control a very complex phenomenon, such as that of visual perception. The proposed methodology can be applied as a further control both in the design of new roads and in the maintenance of existing ones and, in particular, in the analysis of special scenarios (for example unusual conditions of weather or traffic) that are not represented by ordinary audits contained in the road standards.