High-Level Modeling of Component-Based CSPs

Raphael Chenouard (1) , Laurent Granvilliers (1) , Ricardo Soto (1)

1 LINA - Laboratoire d'Informatique de Nantes Atlantique

Most of modern constraint modeling languages combine rich constraint languages with mathematical notations to tackle combinato- rial optimization problems. Our purpose is to introduce new component- oriented language constructs to manipulate hierarchical problems, for instance for modeling engineering system architectures with conditional sub-problems. To this end, an object-oriented modeling language is as- sociated with a powerful constraint language. It offers the possibility of defining conditional components to be activated at solving time, declar- ing polymorphic components whose concrete types have to be deter- mined, and overriding model elements. We illustrate the benefits of this new approach in the modeling process of a difficult embodiment design problem having several architectural alternatives.

Domaines: Intelligence artificielle [cs.AI]

Raphael Chenouard, Laurent Granvilliers, Ricardo Soto. High-Level Modeling of Component-Based CSPs. 20th Brazilian Symposium on Artificial Intelligence, Oct 2010, São Bernardo do Campo, Brazil. pp.233-242, ⟨10.1007/978-3-642-16138-4_24⟩. ⟨hal-00542405⟩