ABSTRACT
One of the most important issues in traffic regulation and control is how drivers behave when they are on the road and how they react to specific prescriptive signals. It is useful to have models which allow us to evaluate, in detail, the number of violations which may occur as a function of environmental conditions. This chapter presents a study on the multilayered feedforward artificial neural networks with backpropagation learning. Data collected was on traffic flow and reactions to outside conditions where it was easiest, i.e. at all access points to a certain number of intersections with traffic lights, chosen in Milan and Pavia (Northern Italy). The aim of this methodology is to construct models of the relationships between driver behaviour, in particular, incorrect behaviour, and specific conditions in traffic flow, within the environment and the requirements of the highway code.
