Image Navigation  Text Navigation

Rapid Capture and Use of Geometric

Concepts in Early Design

GEMCON

Supervisor: Ken MacCallum and Dr A H B Duffy

Research Student: Xiaohong Guan and David Stevenson

Funding: ESPRC

Project Period: July 1993 to June 1996

Background

Geometric design is the process where the geometric properties of a product are established. At early stages of this process, information and ideas are often vague, subject to radical change at all levels, and alternative solution approaches are examined to obtain the best possible concepts.

Despite the critical importance of the early geometric design stages, there exists little computational support that is suitable. By requiring a precise and specific definition of geometry, that is usually unavailable until a later stage, existing CAD systems reflect an early commitment and trial-and-error approach to design. The most successful medium for pursuing conceptualisation and synthesis is, therefore, still &quto;paper and pencil&quto;: annotated sketches (Figure 1). Thus, existing systems are most successfully used for detailing and analysis.

Click on picture to enlarge

Figure 1. Use of CAD systems in the current design practice.

Research Objectives

GEMCON investigated the provision of computational support for early stage geometric design. The major objective was to establish an approach to an interactive geometric design environment which allows rapid capture of, and natural experimentation with, geometric concepts. The captured concepts could then be used for functional evaluation from multiple viewpoints and for re-use in downstream computer-based design processes. Specifically, this project aimed to:

• Develop computational schemes for uniform representation of approximate geometric design concepts at multiple levels of abstraction and for reasoning about the concepts.
• Investigate means for rapid input and presentation of such concepts.

Research Methodologies

A combined theoretical and experimental prototyping approach has been adopted in the research. The theoretical work concentrates on the development of a representation for modelling the geometric concepts and on related reasoning techniques. Experimental prototyping is used to construct prototype systems for modelling and for graphical concept capture.

Research Achievements

A geometric design model has been formulated that provides a basis for the geometric design system being built. A minimum commitment modelling and an incremental refinement principle have been proposed for computer aided geometric design (CAGD) systems to assist in the visualisation and maintaining of the geometric design space implied by a vaguely defined model. Existing applications of the minimum commitment principle have been examined. Implications of adopting these principles in CAGD systems are being explored through their application to the development of a prototype system (GEMCON system).

A number of significant elements involved in modelling and maintaining approximate (and precise) geometric models - thus the corresponding design space - incrementally in multiple levels of abstraction (figure 2), have been identified, and a representation framework and constraint-based reasoning mechanism are being developed to support these. Three areas - modelling, evaluation and conservation - in a CAGD system have been identified as able to benefit from the use of geometric abstraction. The role of geometric constraints as a "natural" language for specifying approximate and precise geometric information has been recognised, and a classification structure has been established.

Discussions with the industrial partner of the project have revealed that, although tolerance information is of least concern at the early design stages, incorporating it into a corresponding support system would help designers develop an earlier awareness of tolerances and facilitate the use of standard and past design components. Issues related to incorporating tolerances into the GEMCON system have been studied. The required representation for recording size tolerances has been established.

A pseudo-3D sketcher has been identified as a suitable basis for building a front-end user interface tool for the underlying geometric application. It seeks to provide for rapid input, presentation and manipulation of application concepts, within the familiar interaction model of a typical 2D drawing package. It has two characteristics: custom-built, which supports for continuing exploration and evaluation of different interaction and presentation techniques, and linked, which supports experimentation in producing both pseudo-objects such as air-flow arrows and non-objects such as annotations.

The software implementation environment of the project is Harlequin CommonLisp/CLOS.

Click on picture to enlarge

Figure 2. GEMCON research focus.

Relevant Publications

Please go to publications page or use search box to find details about Ken MacCallum, Dr A H B Duffy, Xiaohong Guan and David Stevenson.

For More Information Contact: Dr A H B Duffy