Contact: Anne Watzman For immediate release: (412) 268-3830 Jan. 19, 1996 Carnegie Mellon Researchers Receive $2.2 Million To Develop Architecture for a Table-Top-Sized Robotic Factory PITTSBURGH--Carnegie Mellon University researchers have received $2.2 million from the National Science Foundation's Multi-Disciplinary Challenge Program to develop a distributed architecture that will enable the speedy design, set up and deployment of mini-factories operated by robots. To illustrate this concept, researchers will develop the software and hardware to run a table-top-sized factory in which several reconfigurable fist-sized robots will assemble small, high-value products. The initial applications will be to improve the assembly of magnetic disk drives, which will be supplied by Hewlett-Packard Co., and wearable computers which university researchers have developed with support from the Advanced Research Projects Agency. But the concepts developed over the grant's four-year span could enhance many products with significant mechanical and electronic aspects. The project's principal investigators include Ralph Hollis, senior research scientist at the university's Robotics Institute, Mahadev Satyanarayanan, a professor of computer science who is an expert in distributed systems, and Mark Kryder, director of Carnegie Mellon's Data Storage Systems Center (DSSC), the nation's premier university-based research program in magnetic and optical recording technology. "By combining high performance computing, modular robotics and the latest communication technologies, we hope to develop a prototype for designing, setting up and programming factories that will give U. S. manufacturers an edge in responding to rapidly changing global market conditions," says Hollis. The group plans to reduce the time it takes to design and deploy an automatic assembly system from four months to less than a week, increase manufacturing precision from today's 50-100 micrometers to one micrometer, increase assembly flexibility so a factory can change the products it makes during an eight-hour shift and improve the portability of an assembly system by decreasing its size to one-tenth that of a conventional assembly line. Hollis envisions a factory composed of modular robots, all equipped with vision and the ability to network with each other. Such a facility ultimately would be designed by manufacturers and geographically distributed robot vendors who would work together over the Internet. Once the design is finished, the manufacturer would simulate building the product, and depending upon which robotic components work best, access those models at various vendor locations. Vendors would be alerted by e-mail to send the necessary equipment to the manufacturing site by the next day. There, parts would be snapped together with a minimum number of tools and be up and running in a matter of days. The idea of a rapidly reconfigurable factory in which you could convert a drive line from one product to another almost overnight could reshape the economics of the disk drive industry, says Kryder. "Despite its huge size--about $25 billion--and rapid growth, the industry suffers from poor earnings because companies can't change their manufacturing processes fast enough to keep up with the changing marketplace." "Today, a significant opportunity exists to produce many important benefits for the world of manufacturing by applying new computer and communication technologies," adds Satyanarayanan.. "This project is the first step toward distributed manufacturing." The National Science Foundation's Multi-Disciplinary Challenge Program is a component of the Federal government's High Performance Computing and Communications Program. It funds projects directed at fundamental problems in science and engineering with broad economic and scientific impact whose solutions require the application of high-performance computing. The grant shared by Hollis, Kryder and Satyanaraynan is the major element of a larger architecture for an agile assembly project run by Hollis. Its other aspects deal with real-time networking for automated assembly systems, and sensing and control of planar linear motors to enhance their precision. ### 96-31