Shape Synthesis Design Automation from Feature - Bond Input.

We propose a new paradigm for CAD which aims to support the early stages of mechanical design well enough that designers axe motivated to use the workstation as a conceptual design tool. The basic approach is that of shape synthesis - the automated generation of part designs. Many mechanical components can be defined by two kinds of geometry: features which are critical to its function (application features), and material which constitutes the body (bulk shape). Application features are high-level entities in terms of which the designer can reason about the design. Bulk shape must obey certain constraints - such as non-interference, minimum structural strength, etc. Our team has developed a system wherein the designer inputs the application features, along with topological and other constraints, and the bulk shapes of the parts are automatically synthesized. Overall economy is enhanced by reducing the input necessary from the designer, providing a more complete exploration of design space, and enhancing manufacturability of the component parts. We have already developed a simple software system for 2-D component designs. Our Phase I effort will focus on developing this technology for more general design problems. Specific tasks include (1) development of skeleton generation algorithms, (2) skeleton optimization, (3) material synthesis algorithms, (4) user interface development, (5) manufacturing module extensions, and (6) multiply parameterized feature types. Phase II will concentrate on softwaxe development of the prototype into a full-featured CAD design package, with applications to real-world mechanical design problems. University of California, Berkeley (Prof. Sastry and students), a pioneer in this field, will provide sub-contract support during Phases I and II of this project.