A context-sensitive model of driving behaviour and its implications for in-vehicle safety systems

The different models of driver behaviour can be categorized as 'Descriptive' models that focus on what the driver does and 'Functional' models that focus on why the driver behaves the way he does and from that to predict drivers' performance in different situations: demanding situations that elicit peak performance capabilities and routine situations that elicit typical behaviour. The optimal approach might be a hybrid that extracts the most useful features of each. Recently, a variety of driver support and information management systems have been designed and implemented to improve safety and performance. To predict the impact of these systems on driver behaviour, we need predictive models of driver-vehicle-environment interactions. The aim of the European ITERATE project is to develop and validate a unified driver-vehicle-environment (DVE) model. A critical review of existing models led to the identification of the most relevant parameters and variables that need to be included in such models. The selected driver characteristics (and variables used to measure them) are culture (country), attitudes/personality (sensation seeking), experience (hazard perception skills), driver state (fatigue), and task demand (workload). The proposed model includes selected environmental parameters that are simulated in the different test phases, such as road, traffic and visibility. The model will consider driving behaviour and performance from the point of view of how drivers perceive and attend to environmental situations, make choices, and respond to those situations. Performance will be measured in terms of errors and reaction times. Though this paper focuses on cars, the ITERATE project covers trains and ships as well.