Good morning. When I worked at Autodesk, I gave numerous sales presentations in front of admirals, contracting officers and CAD users poised to purchase hundred-thousand dollar CAD systems. Prior to the presentation, a sales representative would make a glowing and fabulous introduction for me. Unfortunately, I can't repeat any of those accolades without offering you nausea medicine so let me just introduce myself Andy Dong and my topic A Mechanical Design Language Interpreter under the direction of Professor Agogino and the Berkeley Expert Systems Technology laboratory.
Transition: I'd like to motivate my research by relating my experience in the CAD industry. Having had the opportunity to work with Autodesk and Sun Microsystems to deliver concurrent engineering systems based on industry standard design drafting programs such as AutoCAD and Pro/ENGINEER, I have found it interesting to note how quickly the industry buys into the latest and greatest CAD technology. The fact is, industry views CAD as the backbone of product design not only for drafting but also analysis, optimization, and decision support.
Yet, even with the technological achievement of CAD programs, the complaint I hear most from my customers is the lack of features in their CAD applications which address the problems of how much time they spend absorbing design information, keeping up with design changes and reconciling problems. If only there existed CAD programs "smart" enough to learn and understand the design so that they can present relevant design information on demand, engineers could be far more productive than any increase in MIPS and MFLOPS or GUI-fied feature in a CAD application.
Transition: In the field of intelligent manufacturing systems, for example, there exist techniques for learning how to manufacture a product by extracting the shape features from the geometry of a design.
| BEST Lab | Department of Mechanical Engineering | University of California at Berkeley