Autonomous vehicle navigation with semantical modelling: A project overview

In this paper we consider the problem related to the autonomous navigation of a robotic vehicle by modeling and describing the semantics of the context associated to the navigation. Considering the semantical modelling of the navigation context, as obstacles, pedestrians, roundabouts, information on the vehicle itself (load, power, etc.), on the robotic vehicle, it is integrating to add this symbolic artificial intelligence to a classical navigation control scheme. This semantic modeling generates an additional degree of abstraction by "reasoning" about the information collected by the various sensors of the autonomous vehicle in order to build a navigation context. This navigation context is able to provide decision support for the autonomous or semi-autonomous vehicle (interaction to the driver). In this paper, we present a project developed at Université de Technologie de Compiègne, Laboratory Heudiasyc (Heuristics, Diagnosis and Complex Systems), in France, showing initial modeling results by defining and selecting the relevant information in the creation of a context for the vehicle, structuring the contextual information so as to be able to reasoning on it and thus dynamically update the context in which the vehicle operates; maintaining information by choosing suitable data persistence models in order to be able to query previous simulations efficiently.