Open Competition 2 - Information Technology

A Distributed Agent-Based Lookahead Strategy for Intelligent Real-Time Decision-Making in Manufacturing

Sponsor: Lookahead Decisions, Inc.

• Project Performance Period: 10/1/2001 - 9/30/2004
• Total project (est.): $2,800,000.00
• Requested ATP funds: $2,000,000.00

To meet the demands of a dynamic manufacturing environment, managers often need to make rapid decisions -- on the scale of minutes -- in response to unexpected customer demands, machine malfunctions, employee unavailability, or schedule changes. Research suggests, however, that such human decisions are made blindly as much as 80 percent of the time. Lookahead Decisions proposes to develop an intelligent decision-making system for automated manufacturing facilities that uses advanced search methods that "look ahead" down chains of consequences -- techniques similar to those used by chess programs -- to make better real-time decisions based on data from the factory floor. With support from the ATP program, the company plans to develop a system that is able to model multi-machine transfer lines with complex resource constraints and autonomously guided vehicles for materials handling. This system will be able to learn the model's parameters from factory-floor data and learn from experience. The project also includes developing a user interface as well as an interface to standard factory information systems, and testing the system at two sites. The software will be developed using standard C++ object-oriented programming in a PC-Windows/NT environment. Participation by Siemens, which is providing a test site, will help ensure that the end system will have a market and channels for distribution. In addition to its application to manufacturing, the systems hold promise for application to medical decision-making, transportation and telecommunications network routing, political decision-making, investment planning, and military planning. The market for reactive decision-making software for manufacturing, currently untapped, could rise to 20 percent of the $8.3 billion manufacturing software market. This class of software would be expected to have the most impact on contract electronics manufacturing, a $100 billion segment of the manufacturing market. By lowering inventory holding costs, such software could save a large company with $60 miillion in inventory $1 million annually in inventory holding costs.

For project information:

Dr. Armand Prieditis, (510) 665-1357
prieditis@lookaheaddecisions.com

ATP Project Manager

Jack Boudreaux, (301) 975-3560
jack.boudreaux@nist.gov