Elsevier

Computer-Aided Design

Volume 37, Issue 10, 1 September 2005, Pages 1081-1093
Computer-Aided Design

Geometry-based semantic ID for persistent and interoperable reference in feature-based parametric modeling

https://doi.org/10.1016/j.cad.2004.11.009Get rights and content

Abstract

In current feature-based parametric design systems, the reusability principle is not fully supported as it was expected. Unpredictability and ambiguity of models often happen during design modification within one system as well as among different systems. This reference deficiency significantly reduces the power of feature-based parametric modeling, where geometry re-evaluation generates unexpected shapes. In this paper, a sufficient condition of B-Rep variance based on geometry continuity in parametric complex Euclidean (PpC3) space is proposed. Shape and relation parameters are differentiated in PpC3, thus parametric family can be defined. A semantic id scheme based on continuity of geometry is developed to solve the problem of naming persistency and to improve interoperability of CAD feature modeling. Hierarchical namespaces localize entity creation and identification. All geometric and topological entities are referred uniformly based on surface ids, and topology semantics is retained in id itself.

Introduction

Feature-based parametric modeling has advanced in the last 15 years and rapidly become the mainstream of mechanical design. Nevertheless, parametric modeling systems share a number of shortcomings related to design process [1]. For example, lack of common definition of features introduces extra barriers in model data exchange and sharing; chronological dependency of features in the modeling process reduces the flexibility of modeling sequences; sequential feature evaluation would easily generate unstable and unpredictable geometry; ambiguous and inconsistent models would be created within one system as well as among different systems. Difficulties of design data sharing and reuse still exist in current systems.

The above modeling procedure related problems are all connected to one basic modeling issue: the semantics of feature is not captured actively and maintained throughout the modeling process using current way of feature specification. Currently, features and associated parameters are defined based on evaluated boundary geometry. The design history and the reference structure are constructed based on geometric elements evaluated in previous steps. The chronological dependency between features is volatile during the design process. As a result, feature interaction affects the interpretation of features.

Lack of common standards of feature definition and representation prohibits extensive design exchange and data sharing. Features in different systems are created in proprietary formats. Interoperability during design collaboration and data exchange can only be achieved at the pure geometry level using neutral formats. Comprehensive information of design features and parameters is not transferable within a heterogeneous environment. Interoperability at the level of feature semantics is needed to enable the reuse of original models.

The primary symptom of lack of semantics in feature specification is persistent naming problem, which is common in current commercial parametric modeling systems. In these systems, evaluated topological entities (e.g. faces, edges, and vertices) in a Boundary Representation (B-Rep) model could be references to new features. The change of a feature may directly affect the features that have reference dependency on it during the model re-evaluation. Some features at later steps may refer to a different entity unexpectedly, or even cannot find the reference. Thus, an unexpected geometry is generated.

A typical example (as in Pro/Engineer®) is shown in Fig. 1(a), where a part is constructed by a protrusion and a circular cut feature, followed by a hole feature. The position of the hole is partly determined by the distance s from the center of the hole to edge e1, which is generated by the cut. If the distances from the center of the cut to its references are changed, by either from b to d horizontally or from a to c vertically, as shown in Fig. 1(b) and (c), respectively, the distance reference of the hole to e1 will jump to edge e2. This is because the id of the edge e1 was assigned to edge e2 after the Boolean operation of the cut, and the orientation information of edges is also used in the re-evaluation process. This naming problem produces unstable and unpredictable parametric models in variational design.

Feature evaluation inconsistency also exists among different systems. Lack of feature semantics forces current systems to use heuristic approaches to handle naming issues. This creates ambiguity and unpredictability between CAD systems in feature re-evaluation. For example, Fig. 2(a) shows a part that is constructed by two protrusions and one circular cut feature, where the location of the second protrusion is constrained by distances a and b. At the same time, the position of the cut is defined by distances c and d. If the values of a and b are set to zeros, the desired design should be the one in Fig. 2(b). However, in current systems, the cut would jump unexpectedly because of the face mergence. The updated part in Solidworks® is shown as in Fig. 2(c), and the new evaluation in Autodesk Inventor® is shown as Fig. 2(d). Both are very different from the intended one.

The direct impact of naming persistency problem is that geometry re-evaluation generates unexpected shapes. The original principle of knowledge reuse and ease of modification in parametric design is not followed adequately by current naming and reference mechanisms. The cause behind this is that design intent of feature definition is not captured actively and thoroughly. The geometric meaning of features is interpreted totally based on the references of evaluated topological entities. The semantics of feature is degenerated into references of topological entities and parameter values, as illustrated in Fig. 3. It causes inconsistency and unpredictability of geometric models. This inconsistency significantly degrades the efficiency of feature-based parametric modeling method.

One fundamental issue associated with feature semantics representation is parametric family. A parametric solid model corresponds to a class of solids, but there is no formal definition or standard for what this class is [2]. Without general understanding of this problem, heuristic and incompatible methods of feature definitions are used in CAD industry. While CSG models are globally parameterized, B-Rep models need extra boundary evaluation steps to apply parametric modeling, which causes the complexity of parametric family. It is not clear how to generate and differentiate members of a family, how to describe a family generally, and how to represent and retrieve the common semantics of family members.

Fig. 4 shows a group of parts within one parametric family yet with very different topological constructs. Common descriptions at the semantic level instead of the topological level are needed to categorize models. Until the parametric family is understood systematically, the efforts to allow exchange of parametric representation are likely to remain ad hoc. Exploring the nature of parameterization is an important portion of understanding the semantics involved in feature modeling.

In order to eliminate the flaws of current systems and improve the robustness and efficiency of the parametric modeling method, maintaining feature construction semantics is necessary. In this paper, a geometry-based semantic approach for design feature representation and reference is presented, in which topological entities are named based on geometric construction, and semantics of topology is embedded in ids. In Section 2, related research work is reviewed. Section 3 proposes a sufficient condition for B-Rep variance based on geometry continuity. Based on that, a semantic ID scheme is developed as described in Section 4. Detailed id update and implementation are discussed in Section 5.

Section snippets

Background

Some solutions of the persistent naming issue have been proposed. In the research of E-REP [3], [4], a topology-based naming method is used. New topological entities are named based on the referred old entities during the feature construction. For example, in an extrusion, a new edge is named by reference to the sweeping vertex, whereas a new face is named by reference to the sweeping edge. When model is re-evaluated, new entities should be identified and matched to old entities. The matching

Sufficient condition for BR-variance

To generally define the parametric family of a solid, sufficient conditions for BR-variance are needed. A sufficient condition for BR-variance based on geometric continuity is proposed for a general definition of parametric family. Here, continuity means: throughout a valid parameter range, small changes in a solid's parameter values result in small changes in the geometry of B-Rep. It is difficult to organize variational or parametric families based on topology continuity. While adjacency of

Semantic ID scheme

In the semantic ID scheme, information of construct relation is included in geometric ids, and geometric meaning of identification is in topological ids. Two basic components of this naming system are namespace and topology semantics.

Naming service

Semantic ids need to be assigned and connected to system-dependent identifiers or physical addresses. This process is executed by a naming server, which can be independent of modeling systems. The naming services include binding (assign semantic ids to physical addresses in internal representation), resolution (look up physical addresses from semantic ids), and update (update semantic ids after model re-evaluation). The structure of the naming server is shown in Fig. 12. Through an interface

Conclusion

In this paper, a sufficient condition of B-Rep variance based on geometry continuity in parametric complex Euclidean (PpC3) space is proposed. Shape and relation parameters are differentiated in PpC3, thus parametric family can be defined. A semantic id scheme based on continuity of geometry is developed such that entities are named based on persistent geometry to solve the problem of topology inconsistency in parametric modeling, which aims to improve the robustness and interoperability of CAD

Acknowledgements

This work was supported in part by the Office of Naval Research, US Department of Defense (Grant number: N00014-02-1-0649).

Yan Wang is a Research Assistant Professor at the Department of Industrial Engineering, University of Pittsburgh. He received his BS in Electrical Engineering from Tsinghua University in Beijing, MS in Electrical Engineering from Chinese Academy of Science, and PhD in Industrial Engineering from University of Pittsburgh. He is a principle investigator at the US National Science Foundation Industry/University Cooperative Research Center for e-Design at the University of Pittsburgh. His research

References (13)

There are more references available in the full text version of this article.

Cited by (0)

Yan Wang is a Research Assistant Professor at the Department of Industrial Engineering, University of Pittsburgh. He received his BS in Electrical Engineering from Tsinghua University in Beijing, MS in Electrical Engineering from Chinese Academy of Science, and PhD in Industrial Engineering from University of Pittsburgh. He is a principle investigator at the US National Science Foundation Industry/University Cooperative Research Center for e-Design at the University of Pittsburgh. His research interests include geometric modeling and reasoning, constraint-driven systems, interoperable and secure information sharing in network-centric computer-aided design.

Bart O. Nnaji has a BS in Physics from St John's University with distinction; an MS and PhD in Industrial and Systems Engineering from Virginia Tech, and obtained a certificate of Postdoctoral studies in Artificial Intelligence and Robotics from MIT. He was Professor of Department of Mechanical and Industrial Engineering at the University of Massachusetts at Amherst till 1996. He is currently the Director of the National Science Foundation Center for e-Design and the William Kepler Whiteford Professor on Engineering at the University of Pittsburgh. Professor Nnaji has received 3 honorary doctorates from international universities. He received 1988 Outstanding Young Manufacturing Engineer Award by the Society of Manufacturing Engineering; 1992 Outstanding Young Industrial Engineer Award; He is a Fellow of Nigerian Academy of Sciences; Fellow of the Institute of Industrial Engineers, and Fellow of the Society of Manufacturing Engineers. He was honored with the US Secretary of State's Distinguished Public Service Award in 1995; Distinguished Scientist Award by the World Bank-IMF in 1998; and the Nigerian President national honor-Officer of the Order of Niger (OON) in 2000. Professor Nnaji has served as principal or co-principal investigator on over $40 million research. He has published 5 books and over 100 technical articles. One of his books, Computer Integrated Manufacturing and Engineering, won the 1994 world best text book prize for Manufacturing Engineering. Professor Nnaji is the founding Editor-in-Chief of the International Journal of Design and Manufacturing and also served as the Editor for the Design Department of Institute of Industrial Engineers Transactions on Design and Manufacturing. Professor Nnaji served as Nigeria's Federal Minister of Science and Technology in 1993.

View full text