Prediction of road track paths by a fuzzy model

This research proposes a fuzzy model built to predict the track paths of a motor vehicle on a two-lane road and two-way traffic. The purpose of this study stems from the awareness that the position of the vehicle within the cross-section hardly coincides with the axis of the lane but deviates from this one because of some conditions generally not provided in the road standards. To solve this problem, the authors prepared an instrumented vehicle, measuring many variables during some trials in a rural road located in Sicily, Italy. The resulting database was treated at first with subtractive clustering techniques to correctly size the membership functions and rules without the subjective help of an analyst. Therefore, the fuzzy model was built only on the basis of the measured data, and the result was the position of the vehicle within the cross-section of the examined road. This research can help the analyst to look for the real cause of user's misbehavior even when the variables involved are fully respectful with the road standard to modify or eliminate disturbing elements in the understanding of the road by the users.