Feature-based similarity assessment of solid models

Abstract

This paper presents our initial efforts to develop a systematic approach for assessing the similarity of solid models based on how they will be manufactured. The goal of this work is to develop methods that, given a solid model representing the design of a new product, query a product information database (of solid models, associated manufacturing plans, and related attributes) and identify existing designs with manufacturing plans similar to some reasonable plan for the new design - or useful as a starting point for creation of a new plan for the new design. Our approach is based on the automatic generation (from CAD models) of graph structures that contain manufacturing information (in the form of manufacturing features). We are developing ways to measure similarity among these graph structures, so that given the graph structures corresponding to two different designs, we can tell how similar or different they are. The similarity measure will be used as a basis for indexing and retrieving similar designs from databases. An implementation of our approach is discussed. We believe our work is a first step in producing computer-generatable and computer-interpretable similarity assessment techniques that will be useful for applications such as variant and hybrid variant/generative process planning systems, indexing schemes for large part inventories, access methods for `smart catalogs,' and for performing component searches through product catalogs and on the Internet.