A formal driving behavior model for intelligent transportation systems

Abstract

Vehicular Ad hoc Networks are considered recently as a fertile field of research. Their applications are showing a growing importance as they are expected to improve road safety and traffic efficiency, through the development of vehicle safety applications whose main goal is to provide the driver with assistance in dangerous situations. Thanks to vehicular communications, drivers can permanently receive information about road conditions which help them to make more reliable decisions. The idea behind this paper is to enable an adaptive assistance to drivers in different situations, based on their past driving experience. As a first step, we focus on the modeling and learning of individual driving behavior at a picoscopic level. This paper proposes a formal description of a driver-centric model, using the formalisms of hybrid IO automata and rectangular automata. Then, an online passive learning based approach for the construction of the described model is proposed. Having a model that describe the behavior of drivers can enable us to predict and recognize a driver preferences in different driving context, enabling thus an adaptive assistance. © 2014 Springer International Publishing.