In current product design, significant effort is put into creating aesthetically pleasing product forms. Often times, the final shape evolves in time based on designers’ ideas externalized through early design activities primarily involving conceptual sketches. While designers negotiate and convey a multitude of different ideas through such informal activities, current computational tools are not well suited to work from such forms of information to leverage downstream design processes. As a result, many promising ideas either remain under-explored, or require restrictive added effort to be transformed into digital media. As one step toward alleviating this difficulty, we propose a new computational method for capturing and reusing knowledge regarding the shape of a developing design from designers’ hand-drawn conceptual sketches. At the heart of our approach is a geometric learning method that involves constructing a continuous space of meaningful shapes via a deformation analysis of the constituent exemplars. The computed design space serves as a medium for encoding designers’ shape preferences expressed through their sketches. With the proposed approach, designers can record desirable shape ideas in the form of raw sketches, while utilizing the accumulated information to create and explore novel shapes in the future. A key advantage of the proposed system is that it enables prescribed engineering and ergonomic criteria to be concurrently considered with form design, thus allowing such information to suitably guide conceptual design processes in a timely manner.

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