Modelling Gap Acceptance Behavior of Two-Wheelers at Uncontrolled Intersection Using Neuro-Fuzzy

Abstract

The main focus of this paper is to develop gap acceptance model for two-wheelers using neuro-fuzzy approach at three legged uncontrolled intersections in India. Unsignalized intersections are controlled with the help of stop or yield sign in developing countries, but in India, signs do not work and thus they are treated as uncontrolled intersections by the road users. The processing of vehicles at unsignized intersection is complex and very interactive, whereby each driver makes individual decisions about when, where, and how to complete the required maneuver. Such intersections cannot be analyzed based on the findings of the unsignalized intersections in other countries (HCM 2000, for example). This paper systematically analyzes the behavior of right turning two-wheelers at uncontrolled intersections. Data on gap acceptance behavior is obtained with Video. The various parameters that are collected include vehicle arrival rate, gap accepted/rejected, time to cross the intersection, driver's gender and approximate age, type of conflicting vehicle, and occupancy of two-wheelers. The critical lag and gap by Raff method for two-wheelers is found to be 2.51 and 2.47seconds respectively. This value is smaller than the critical gaps reported in other studies. An Adaptive Neuro-Fuzzy Interface System (ANFIS) has been developed that estimates the possibility of accepting a given lag/gap based on various drivers and traffic attributes. Eight different combinations of attributes are considered. For model calibration, 80% of the extracted data is used and remaining is used for model validation. The percent correct prediction by all models for validation data is found to be significant.