🚀 go-pugleaf

Voordracht Cad/Cam

vrijdag 20 November, 14.00 uur
TU Delft, TWI, zaal C, Julianalaan 132, Delft

The current development from Computer Aided Design towards Computer Integrated 
Manufacturing has pointed out some serious limitations of the current geometric 
models for describing a product. The fundamental data structures of existing 
solid modelers provide information, which is either not sufficiently detailed 
or too detailed. They do not allow the inclusion of non-geometric technological 
information and do not give means for formally representing the functionality 
that is the meaning of the product in the application context. For this reason 
currently considerable attention is being focused on the potential of features, 
viewed as abstract elements that can integrate functionality with geometry 
within the product description. Features, or more properly shape features, 
facilitate the communication between the application and the geometric modeling 
system, since they characterize certain types of shape or refer to portions of 
a part model, which are of interest in the application context. Two approaches 
have been studied for the creation of feature based models: design by features 
and feature extraction from traditional solid models. At CNR a system for the 
automatic feature recognition in multiple functional viewpoints is developed, 
which consists of two modules: a shape feature extractor and a converter to a 
context dependent feature-based model.

Dr. Franca Giannini
Consiglio Nazionale delle Ricerche (CNR), Genova, Italy

Shape Feature Extraction From Boundary Models

In this presentation we will discuss the feature shape extractor, which analyzes
the face-based boundary representation of the product, indentifies the faces 
constituting a shape feature and classifies them as Protrusion or Depression. 
The identified sets of faces are then completed by adding new faces, edges and 
vertices to obtain the volumes corresponding to the identified features. Then 
these volumes are organized in a hierarchical structure called SFOG (Shape 
Feature Object Graph), which is still a geometric model. The SFOG is converted 
to a context dependent feature model by the converter, which is discussed in 
the next presentation.

Teresa De Martino
Consiglio Nazionale delle Ricerche (CNR), Genova, Italy

Application Feature Recognition

The geometric model (SFOG) returned from the shape feature extractor is a 
neutral description, which does not take the application context into account. 
The extracted shape features could be different from the application features 
of interest. The converter performs a user-assisted recognition of application 
features on the SFOG. Using a "teaching-by-example" technique the user can 
define his own features of interest. The system then recognizes instances of 
the feature template in the SFOG through graph matching and user defined 
constraints, which allow different levels of definition for general or specific 
features. In this way the SFOG model can be upgraded to a context-dependent 
feature-based model, ready to be completed with functional and technological 
information.

Inlichtingen: Reinier van Kleij, tel. 015-784564
	      Wim Bronsvoort, tel. 015-782533