RNTI

MODULAD
Extraction et Inférence de Connaissances à partir d'Assemblages Mécaniques Définis par une Représentation CAO 3D
In EGC 2017, vol. RNTI-E-33, pp.21-32
Abstract
Knowledge extraction from 3D geometric models and reasoning is an important issue to the development of ontologies capable of functional description of man-made products. In this context, we rely on an architecture for deductive logic containing a knowledge base coupled with a 3D geometric modeler. Reasoning mechanisms addressing the concept of shape of 3D objects are still difficult to formalize and geometry related information is difficult to extract from 3D models. We propose formal properties such as 'has the same shape', 'belongs to the same family' to show how the extraction of 3D geometry is connected to these properties. Then, formal properties such as 'is a pilling', 'is a grouping' are introduced to show how complementary reasonings can contribute to 3D assembly structuring. These properties are illustrated through an example of hydraulic pump.