WICS'94 Page

WICS'94 Computational Support for Concurrent Engineering


CANCELED - 21 Jul 94
A workshop given by Mark Cutkosky and Charles Petrie for the Western Institute of Computer Science (WICS) and the Stanford University Computer Science Department, August 8-12, Stanford University.
Registration information:
Joleen Barnhill
WICS
P.O. Box 1238
Magalia, CA 95954-1238
barnhill@hudson.stanford.edu

Abstract

The desire to provide computer-aided support for concurrent engineering places special requirements both on the planning and analysis tools that engineers use and on the computational framework that ties them together. This short course will examine a variety of solutions for fulfilling these requirements, based upon experiences with several concurrent engineering systems in different application domains. The right solution depends on the scale and complexity of the problem to be addressed and on the type of design (routine, configuration or novel). Technical issues include task decomposition, design model representation, knowledge sharing and the management of change propagation. Equally important are the human aspects of concurrent engineering. The system must satisfy team members' preferred work styles and help them to achieve a shared understanding of the project.

Issues Covered


Course Outline

Day 1

What is Concurrent Engineering?
We focus on the computational support of design engineers working together to to design products and develop the manufacturing plans needed to produce them.

We give a description of the problem and solution spaces, describing the fundamental theoretical issues, such knowledge sharing and task coordination.

We survey other relevant work, both theoretic and applied and give pointers to other sources of information. This survey is a synopsis of recent publications in conferences, workshops, and books.

We discuss the particular topic of Conflict Management. This is now recognized to be a crucial topic that is not yet well-understood. We describe the current approaches to this problem and what can be done pragmatically.

Day 2

Stanford Concurrent Engineering Systems
We give an overview of the First-Cut, Next-Cut, and FIRST-LINK systems from the Stanford Center for Design Research. These will provide a context for a detailed case study involving a distributed agent-based system called NEXT-LINK.

We describe an experiment called PACT in sharing information among diverse, geographically separated, knowledge-based systems. We also describe the SHADE project and associated tools, languages and representations that grew out of the PACT experiment, including KIF, KQML, and Ontolingua from the Stanford Knowledge Systems Laboratory (KSL). We will also cover the ICM project from CIFE.

We describe the current Next-Link system and its capabilities. Next-Link is a system for design and routing of aircraft electrical cables. We will give a live demonstration of multiple agents and using the Next-Link system to coordinate the design of a cable harness.

We then describe the fundamental issues of representation and design rationales using the Next-Link system as an example. We show that some of the features of this system are advanced but likely to be part of future useful systems in which multiple design engineers cooperate to produce a complex artifact. In particular, we show how this can be accomplished without requiring that each engineer use the same data repository, or even the same model of the artifact.

Day 3

The Redux Theory of Design
We describe the Redux theory of design that is used to coordinate the engineers and tasks within the Redux system. This theory allows heterogeneous systems to coordinate their tasks by sending messages to a Redux agent.

We briefly describe the artificial intelligence concepts employed, including search, dependency-directed backtracking, and truth maintenance. The goal of this part of the seminar is to transfer the concepts sufficiently to allow the participants to reimplement and modify them for their own concurrent engineering applications.

We describe how Redux works by showing examples from the Next-Link system, as well as examples from other applications of Redux.

There will be a student exercise in which the participants will collaborate on a design task using Redux coordination services. This will provide a hands-on opportunity to explore the capabilities of cutting-edge research and gain an early understanding of this future technology.

Day 4

Design and Process Planning
We describe the First-Cut and Next-Cut systems for process planning in more detail. We give a methodology and computational framework for concurrent product and process design that allows design to be incremental and hierarchical, and allows design commitments to be delayed when flexibility is important.

We survey use of artificial intelligence techniques in design and process planning and give examples of expert systems for injection molding and milling machine planning.

Internet Collaborative Tools We describe tools developed by Enterprise Technologies Inc. in collaboration with the Stanford Center for Design Research that take advantage of Internet facilities such as MBONE and World-Wide Web. These include using the Internet for video conferencing, shared whiteboards, asynchronous graphical annotation, prototyping, and product catalogs. We also discuss the state-of-the art for electronic design notebooks. This work is the result of research sponsored by the SHARE project funded by the ARPA MADE program.

Day 5

Agile Manufacturing Experiments
We describe two very recent experiments in providing manufacturing services over the Internet. These experiments are a microcosm of the future of electronic commerce.

One is the development of an agile manufacturing capability called ACaPS. The technology in this system allows customers to from specifications to design to semi-automated manufacture of electrical cable harness, using the Internet, and deliver prototypes in a few days.

The second is a recent ARPA experiment in collaborative design. A missile infrared tracker was redesigned for a demonstration. The significant part of this experiment was that it was performed concurrently by multiple universities and companies. The design and build part of the experiment was performed in in only three months. We describe the experiences of the participants and the lessons learned in this experiment, called MADEFAST.

_____________________________________________________________________
Charles Petrie