Decision Making at a Crossroad: a Negotiation of Contexts

The initial training generally puts the driver in a supervised learning context that is not the autonomous-driving context in which decision making and responsibility are assumed by the driver alone. As a consequence, the beginner is not able to contextualize the learned procedures and transform them in effective practices, to identify and manage pre-critical situations and to evaluate the degree of danger of situations. Starting from the studies and works presented in the literature, we propose a twofold approach based on the one hand on global methods and, on the other hand, on local methods. We first represent a driver in two spaces (situations and behaviors) related through scenarios. The situation space is an objective representation by a lattice and the other one is a subjective representation by contextual graphs. Our second goal is to provide the driver with a system for a self training and self evaluation of his behavior in different pre-critical situations. This paper discusses the cognitive part of the approach and objectives on a case study.