ME394 Computational Support for Team Design
Time: Wednesdays 10:00-11:00am
Location: CDR lobby, Bldg 560
1 unit
Instructor: Cutkosky
and
Petrie
Description
This seminar will focus on modern computer-aided approaches
for supporting teams of designers. The emphasis will be
on supporting interdisciplinary teams, where problems
arise in exchanging information among different specialities
(e.g., electronics design, mechanisms, manufacturing)
and in coordinating the efforts of the various specialists.
Speakers will be drawn mainly from the local research community.
The format will consist of 1/2 hour presentation
followed by 1/2 hour of discussion.
Talk abstracts will be emailed & posted in advance.
For 1 unit of credit, each attendee must prepare a question that
they will ask of the speaker at least 3 times during the quarter.
The list of questions will be compiled by the instructors.
Schedule of Talks
17 Jan 96
"The Next-Link Architecture for Distributed Design"
Charles Petrie
<petrie@cdr.stanford.edu>
Center for Design Research (CDR)
24 Jan 96
"ASCAD: Agent-Supported Collaborative Design"
Yan Jin
<jin@cive>
Center for Integrated Facility Engineering (CIFE)
31 Jan 96
"The Virtual Design Team (VDT)"
John Kunz
<kunz@cive>
CIFE
7 February 96
"Interdisciplinary Communication Medium (ICM)"
Renate Fruchter
<fruchter@cive>
CIFE
14 February 96
"PROTEGE-II: Construction of intelligent systems from
libraries of reusable components"
Mark Musen
<musen@camis>
Knowledge Modeling Group,
Section on Medical Informatics (SMI)
21 February 96
"Building
an Infrastructure for Collaboration over Distance"
Paul Losleben
<losleben@cis>
Center for Integrated Systems (CIS)
28 February 96
Wide Area Document Management
Jay Glicksman
<jay@eit.com>
Enterprise Integration Technologies (EIT)
6 Mar 96
What is an Ontology?
Adam Farquhar
<axf@hpp.stanford.edu>
Knowledge Systems Laboratory (KSL)
13 Mar 96
Topic: Multidisciplinary Optimization
in Aircraft Design
Steve Altus, Ian Sobieski
Aeronautics and Astronautics
See also
Agent-Based Engineering
Abstracts
Next-Link is a framework consisting of a reusuable KQML-based agent
message protocol and reusuable generic agents that can be used with
domain-specific agents in order to coordinate their tasks in an
engineering project. Next-Link is designed to be used with projects in
which the tasks and subtasks and be analyzed prior to execution and
appropriate agent "wrappers" written for legacy CAD systems. This talk
illustrates the domain-independent Redux model of design coordination
used as a basis for Next-Link in the context of electrical cable
harness design.
Our research on ASCAD attempts to develop an intelligent agent-based
framework to support coordination for collaborative engineering
teams. In ASCAD, a collaborative design team is a collection of
design-cells and each design-cell is composed of a designer, a
software agent, and a number of computer tools. The role of an agent
is to help its designer coordinate with other design-cells: recording
design and inter-designer interaction process information as part of
"design context", identifying needs for coordination, sending "request
for coordination" to relevant design-cells, establishing direct links
between designers, and providing information and suggestions for
coordination decision-making. This lecture will discuss the research
issues, theoretical framework and current status of the ASCAD project.
Businesses have started to "reengineer" their processes and
organizations. Increasingly, they use client-focused projects rather
than functional organizations, outsource many components and services
to best-in-class providers, reduce the amount of middle management,
increase use of communication tools, and redefine tasks to emphasize
adding value for the end customer. Businesses now reengineer their
operations using best-available judgment, without the ability to make
detailed predictions of the potential effects on business objectives
of changes in organization structure or support systems.
The Virtual Design Team (VDT) research group has developed computer
tools and methods to analyze the effectiveness and limits of
alternative ways to reengineer an organization. These tools enable an
organization designer to start to design an organization the same way
an engineer designs a bridge: using computational tools to predict the
system behavior under many possible demands on the organization. This
talk will discuss a case example of VDT use in the aerospace industry
and introduce the underlying theory and methods.
The objective of this project is to develop a computer environment
which will improve the communication among designers in an
interdisciplinary team. We present a
Propose-Interpret-Critique-Explain (PICE) paradigm as the
communication cycle for collaborative conceptual design. We explore
and test the PICE conceptualization by modeling it with a software
prototype,
ICM , that integrates
graphic representations of a shared 3D model of and AI reasoning tools
about, the evolving design. ICM provides a graphic environment as
the central interface to reasoning tools, to support design. The
graphic environment enables designers to explore the toplevel
functional object definition of the future device, as well as the
AutoCAD form model of the device. The goal of the ICM prototype is to
help improve the quality of design by supporting: (1) improved
concurrent engineering, (2) increased number of explored alternatives,
(3) multi-criteria evaluation, (4) reduced product design cycle, (5) capture of design intent,
(6) smooth electronic transition to later stages of product development.
The notion of sharing and reusing the knowledge encoded within
knowledge-based systems is receiving considerable attention in the
artificial-intelligence communtity. Most work to date has
concentrated on development of standards for declarative knowledge
representation. The engineering of large-scale knowledge based
systems, however, demands attention not only to representation of
logical propositions about the world being modeled, but also to the
control knowledge that allows complex problem solving to take place,
and to the cognitive requirements of building large knowledge bases.
For the past several years, our research group has been building a
development environment, known as PROTEGE-II, that facilitates reuse
of knowledge by teams of developers in multiple ways. PROTEGE-II
supports libraries of reusable problem-solving modules that define--in
domain-independent terms--the manner in which propositional domain
knowledge may be used to solve application tasks. PROTEGE-II also
allows system builders to create and edit domain models--which
themselves may be reusable--and to map those models to the knowledge
requirements of the problem-solving modules in well defined ways. The
result is an architecture that offers system builders the ability to
develop knowledge-based systems from reusable building blocks.
Furthermore, PROTEGE-II processes domain models to generate
automatically custom-tailored knowledge-acquisition tools that
application experts can use independently to enter the content
knowledge required to define individual application tasks.
Our experience in building a variety of large knowledge-based systems
demonstrates the utility of our resuable building blocks, and the
benefits of separating the creation of abstract models by system
analysts from the instantiation of those models with content knowledge
by domain experts.
How can we use networking technology to make a geographically
dispersed community of researchers more productive? This presentation is a
progress report on a project at Stanford which seeks to accomplish this
objective for a specific discipline--semiconductors. The emphasis in this
project is on building practical solutions that work in a very conservative
discipline. Topics to be covered include electronic publishing, real-time
collaboration tools, remote resource sharing and a few examples of novel
new ideas.
Teams produce and need to share many documents in the course of their
work. We use "document" in a very loose sense and include design
documentation, Web pages, CAD files, 3D image files, etc. Any
information that can be represented as a file on a computer needs to
be managed throughout the design process. We are developing a system
for distributed teams to manage documents over the Internet whose
major functions supported are access and version control. In addition
administrative capbilities are being implemented to manage the
overlapping spaces of users, groups, documents, and sessions.
The term "Ontology" has become increasingly widespread in recent
years. In this talk, I will discuss problems that arise when
different people, groups, and systems communicate and how explicit
ontologies can help to solve them. Rather than providing a formal
characterization, I will also compare and contrast ontologies with
glossaries, databases, database schema, class libraries, knowledge
bases, and so on. I will discuss concrete examples of how ontologies
can be used and what they contain. This should help to build up our
intuitions about what an ontology is and is not. I also hope to
clarify the novel and challenging aspects of research on the design,
construction, sharing, use and reuse of ontologies.
The development of an aircraft is a computation-intensive design
problem, involving many individuals and organizations. Large-scale
multidisciplinary optimization problems, such as those encountered in
aerospace design, pose difficulties for conventional optimization
architectures. This talk describes some of these difficulties,
summarizes strategies for dealing with them, and suggests areas for
continued research. We will describe ideas for analysis and design
decomposition and current research on collaborative optimization.
Related pages are:
WWW
Mechanical Engineering Virtual Library
DesignNet - a resource for engineers and designers
Agent-Based Engineering (ABE)
Charles J. Petrie
Stanford Center for Design Research
<petrie@cdr.stanford.edu>