An Adaptive Neuro-Fuzzy Inference System for Simulation of Pedestrians Behaviour at Unsignalized Roadway Crossings

“Gap acceptance” behaviour oversees pedestrians crossing manoeuvre at unsignalized road crossings. From a scientific point of view, the study of pedestrians behaviour has a particular interest, since the underlying factors of behavioural interaction between pedestrians and motor vehicles drivers have a strong non-deterministic component, which makes their simulation very complex. In this paper a Fuzzy logic model for representation and simulation of pedestrian behaviour in such a manoeuvre is proposed. The calibration of Fuzzy model membership functions is executed through an Adaptive Neural Network which considers a sample of “gap acceptance” decisions collected on field. The analysis method is at first theoretically defined and then applied to a real pedestrian crossing.