Toward a modelization of drivers behaviour in different traffic situations for the design and the evaluation of advanced cooperative systems

Toward a modelization of drivers behaviour in different traffic situations for the design and the evaluation of advanced cooperative systems This paper presents the first experiment performed in the frame of a French research national program in the ground transportation: ARCOS. The aim of this study is the development of an advanced cooperative system able to modify its parameters in function of the driver’s behaviour. The first stage of this study consists in the building of an experiment which aims to determine the driver’s behaviour during typical traffic situations without assistance tools and, from this study, to determine the variables able to differentiate various classes of comportment during an overtaking. These experimentations have been carried out on SHERPA driving┬ásimulator. 39 subjects had to drive on a combination of A-roads and B-roads, with various traffic situations. During these experimentations, drivers were filmed and variables, characteristic of vehicle position in relation to the road and of the driver’s actions on the vehicle, were recorded. The Multiple Correspondence Analysis (MCA) was applied on┬átheses┬ávariables, before hand cut into space modalities, to classify driving behaviour during different traffic situations and to identify the best set of recorded variables that allows to discriminate those classes. This article presents results of a study on the first part of a “classic” overtaking. The MCA results show that three variables are able to classify all drivers in function of their driving behaviour during this phase. In the future, the analysis of the others situations will allow to define behaviour styles of the drivers in function of traffic behaviour.