Applying a Human Centred Design (HCD) approach to Intelligent Transportation Systems (ITS) is a challenging issue. Indeed, it is fundamental to confront the end-user to the designed system as often and as soon as possible. However, this process is both costly and time-consuming. We present several ways in which driver models, especially cognitive models, could support the HCD of driving assistance systems. As developing and implementing a cognitive model is a very challenging issue, we propose an iterative method to develop such a model. This method relies on the cognitive framework COSMODRIVE and on specific Knowledge Engineering tools.