From Inspiration to Creation - Extraction and Reuse of Shape Characteristics to Support Product Design

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

This award provides funding for the development of a computational technique to allow the identification and reuse of the characteristic shape identities of products from their computer models. The developed algorithms will extract styling information embodied in the geometry and make its elements available for use in future products. This will be accomplished using an identity-preserving hierarchical shape abstraction technique that allows 3D models to be represented with progressively simpler variants involving a suite of base and detail geometries. The identified characteristics will serve as manipulable shape representations comprised of volumes, surfaces, and curves. The algorithm will be implemented and deployed in a geometric design software that allows individual pieces of the deciphered geometries to be assembled, transferred, and blended in innovative ways to facilitate the creation of novel product forms. If successful, the results of this research will lead to new computer-aided design methods that enable inspirational past designs to inform future designs. The developed techniques will help designers concurrently study large groups of models to reveal historical trends and make future projections. It will also allow companies establish brand identities much more effectively, thereby enhancing style consistency across product families. The shape knowledge computed using the new approach will allow the development of quantitative metrics to measure the strength of analogy between designs. This capability will advance the boundaries of future search algorithms, allowing analogy-based descriptors to facilitate 3D shape retrieval. The outcomes of this award will also enhance design education by allowing experts to better illustrate their thought processes to others, and to quantitatively guide and assess trainees.