Conceptual design is the initial stage of the transformation between functional and structural descriptions of devices. Our primary aim is to develop ideas that will allow computer tools for conceptual design to be built. In this paper, we point out a fundamental trade-off between the expressiveness of design languages and the computational complexity of the resulting design problem. Research in computational conceptual design can therefore be viewed as the problem of defining a design language and then devising ways of controlling the size of the resulting design space. We propose that an effective means of controlling complexity is to use knowledge of existing designs to guide the synthesis of novel designs. We illustrate this concept with a program that designs novel mechanical fasteners from knowledge of existing fasteners. We analyze this experiment and highlight several areas for future work.

This content is only available via PDF.
You do not currently have access to this content.