|
Research and Engineering of Intelligent Systems
Developing scientific and engineering foundations for
metrics and standards of intelligent system
Need
Intelligent control is a critical enabler for realizing
increases in productivity and quality during production. The Integrated
Manufacturing Technology Roadmapping Initiative (IMTI) and Association
for Manufacturing Technology (AMT) roadmaps call for capabilities such
as self-diagnostics, adaptive control, error compensation, thermal and
load compensation, and tool wear and breakage monitoring. AMT predicts
increases in productivity of individual machines by an order of magnitude
by 2010, while increasing machining accuracy by a factor of 5.
All of these functional advances in machine control require
intelligent control. Controls at the individual machine, the unit process
level, today can take symbolic commands and repetitively produce commanded
motions. To achieve the advances identified above requires adding the
capabilities of real-time sensing, of understanding the process to be
controlled, and of predicting future behavior so as to always be able
to take correct action even when encountering unexpected situations.
Everyone refers to the terms "intelligent systems" and
"intelligent control" but these terms are not well defined or understood.
The Institute of Electrical and Electronic Engineers (IEEE) Task Force
on Intelligent Control catalogued a multiplicity of concepts of non-classical
control which are variously called "intelligent" and which in many cases
are at odds with each other in philosophy, in approach, and in application.
Clearly there is a need in this research community for quantitative
and objective performance metrics to define and structure the problem
domain, to improve communication among researchers, and to lay the groundwork
for describing and evaluating products in future competitive markets
when the research is reduced to technology and commercialized.
The near term measurements and standards needs in research
and engineering of intelligent systems are then: performance metrics
for components and systems, data standards for knowledge engineering,
and architectures and interface standards to enable adding "intelligence"
to machine controls.
Goals
To develop the scientific and engineering foundations
for metrics and standards of intelligent systems.
Approach
In response to the needs identified above, the Research
and Engineering of Intelligent Systems Program addresses the solution
of four problems of significance to robotics and manufacturing control
in general.
1. Metrics. Thousands of institutions are working on intelligent
systems, but without a definition of what that means in a quantitative
sense. There currently is no way to measure performance or compare research
results. Therefore the need is for quantitative metrics which can be
used to assess the performance of a system or of its sub-components.
2. Knowledge Engineering. Intelligent control is usually
grounded on knowledge based perception or control. Most knowledge engineering
techniques, such as ontologies, are not suitable for real-time applications.
Therefore, there is a need for engineering tools for building knowledge
bases suitable for real-time perception and control. Open architecture
controls allow the possibility of running user code and real-time process
models; this program objective addresses how to build those real-time
models.
3. Architecture and Tools. Realizing product and quality
gains of intelligent control requires a seamless flow of information
throughout the manufacturing process. Such interoperability in complex
systems requires agreement on components and their interfaces, namely
on the architecture. Therefore the need is for infrastructural definitions
for building and integrating open-architecture based intelligent systems,
such as architectures, principles, standards, and generic tools for
designing and building complex systems.
4. Learning. One of the key technologies required to fulfill
the anticipated benefits coming from intelligent control is learning.
The need is to extend NIST's Real-Time Control System (RCS) architecture
to include learning, self-optimization, self-diagnosis, and adaptive
control, so that we may better serve industry in developing principles,
metrics, and standards pertaining to these technologies.
Benefits
The market that would benefit from this work is significant.
Annually, the U. S. machine tool market is over $5 billion, the U. S.
robotics market is $1 billion, and the inspection market is over $500
million. These industries in turn leverage the discrete part manufacturing
industries that represent almost $2 Trillion of the gross domestic product
(GDP) and are the principal players in exports and in defending against
imports. Savings of even a few percent represent a huge payback to U.
S. manufacturers and the U.S. economy as a whole.
Progress
Conducted a workshop, jointly with IMTI, on the topic of "First Part
Correct." Work has begun on the high priority issues that emerged from
the workshop. The roadmap resulting from the meeting is being published.
Created an integrated inspection testbed demonstration, showcasing feature-based
process planning, component-based software development for controller
nodes, and web-capable visualization of inspection results. This testbed
is being used to study architecture, software development, and interfacing,
and performance issues associated with open architecture intelligent
controllers. This work created the foundation for the MAA workshop
and follow-on interfacing, standardization, and validation activities
now underway jointly with industry groups.
Co-organized, with Metrology Automation Association, a workshop on Open
Architectures in Metrology Automation. Published proceedings and created
an action plan to address issues raised at workshop. MAA members urged
MEL to provide leadership and resources to help their industry segment
attain open architectures in metrology equipment. Extensive follow-on
work is underway, carried out by the Intelligent Open Architecture Controls
Program.
Initiated a series of workshops on Performance Metrics for Intelligent
Systems, which bring together researchers and practitioners from multiple
disciplines to share their perspectives on metrics and machine intelligence
and to prioritize research directions. PerMIS 2000 and
PerMIS 2001 were very succesful and the 2002 workshop is being planned.
Developed the de facto international reference test arenas for
Urban Search and Rescue (USAR) robot competitions. This unique
test course, which represents one particular approach to measuring "intelligence,"
provides a simulation of a collapsed structure, through which robots
have to navigate and locate potential human victims and hazards and
report back to humans. Initially fielded at the American
Association for Artificial Intelligence (AAAI) 2000 conference, the
arenas have been adopted by the RoboCup Rescue organization.
In 2001, the arenas were used for the joint AAAI mobile robot and RoboCup
Rescue competition. RoboCup Rescue will leave behind an arena
in each country where the competition is held, leading to a world-wide
dissemination of the same arenas. This work is partly funded
by the DARPA Mobile Autonomous Robot Software Program.
Concepts in hierarchical world modeling and cost-based planning originally
developed within the research program proved highly succesful when adopted
and enhanced for the Demo III autonomous scout vehicle.
Experimented with several representation techniques for real-time architectures
and component specifications. These include Architectural Description
Languages and the Universal Modeling Language (UML).
An SBIR was awarded to Pathway Technologies, Inc. for work on a simulation
and animation environment for building intelligent controllers, based
on the RCS architecture. SBIR received a Phase II award to continue
developing their toolset.
Received multi-year funding from DARPA Mobile Autonomous Robot Software
Program to develop means of estimating requirements for building truly
autonomous on-road driving vehicles. Project thrusts include
development of overall architecture, performing task decomposition,
building simulation systems for experiments, evaluating sensor requirements,
and developing performance measures. For FY02, we are collaborating
with the Information Technologies Laboratory to evaluate Human-Robot
Interaction.
Significant recent publications include:
Research and Engineering of Intelligent Systems Publications
and Talks
Get
Adobe Acrobat Reader
Jacoff, A., Messina, E., Evans, J., "Performance Evaluation
of Autonomous Mobile Robots," to appear in Industrial Robot 29:3, May
2002.
Lacaze, A., "Hierarchical real-time path planning for one or more autonomous
vehicles," Proceedings of SPIE Vol. #4715, April 2002.
Balakirsky, S. and Lacaze, A. , "Value-driven behavior generation
for an autonomous mobile ground robot," Proceedings of SPIE Vol. #4715,
April 2002.
Tsai H., Balakirsky, S., Messina, E., and Shneier, M., "A hierarchical
world model for an autonomous scout vehicle," Proceedings of SPIE Vol.
#4715, April 2002.
Meystel, A. and Albus, J., Intelligent Systems: Architecture, Design,
and Control, Wiley, 2002.
Messina E., Evans J., Albus, J. "Evaluating Knowledge and Representation
for Real-Time Control," Proceedings of 2001 Performance Metrics for
Intelligent Systems, Mexico City, Mexico, September 2001.
Jacoff, A., Messina, E., "Experiences of Deployment of Reference Test
Arenas for Autonomous Mobile Robots," Proceedings of 2001 Performance
Metrics for Intelligent Systems, Mexico City, Mexico, September 2001.
Albus, J and Meystel, A., Engineering of Mind: An Introduction to
the Science of Intelligent Systems, Wiley, 2001.
Gazi, V., et al, The RCS Handbook: Tools for Real-Time Control Systems
Development, Wiley, 2001.
Kramer, T., Huang, H., Messina, E., Proctor, F., Scott, H., "A Feature-Based
Inspection and Machining System," Computer-Aided Design, August 2001.
Meystel, A. and Messina, E., Editors, Measuring the Performance and
Intelligence of Systems: Proceedings of the 2000 Performance Metrics
for Intelligent Systems Workshop, Gaithersburg, MD, NIST Special
Publication 970. pdf
version
Huang, H.M., Messina, E., Scott, H., Albus, J., Proctor, F., Shackleford,
W., "Open System Architecture for Real-time Control Using a UML Based
Approach," Proceedings of the 1st ICSE Workshop on Describing Software
Architecture with UML, Toronto, Canada, May 15, 2001. pdf
version
Michaloski, J., Birla, S., Igou, R., Weinert, G., and Yen, C.J., "An
Open System Framework for Component-Based CNC Machines," ACM Computing
Survey Symposium on "Object-Oriented Application Frameworks," March
2000. pdf
version
Albus, J.S., "Features of Intelligence Required by Unmanned Ground Vehicles,"
Proceedings of the Performance Metrics for Intelligent Systems Workshop,
NIST, Gaithersburg, MD, August, 2000. pdf
version
Meystel, A., and Messina, E., "The Challenge of Intelligent Systems,"
Proceedings of the 2000-15th IEEE International Symposium on Intelligent
Control, July 2000. pdf
version
Messina, E., Huang, H., Scott, H., "An Open Architecture Inspection
System," Proceedings of the 2000 Japan-USA Symposium on Flexible Automation,
July 2000. pdf
version
Horst, J. A., Messina, E., Evans, J. M., Swyt, D. A., Editors, Proceedings
of the Open Architecture in Metrology Automation Workshop,
Gaithersburg, MD, May 2 & 3, 2000. pdf
version
Messina, E., Dabrowski, C., Huang, H., Horst, J, "Representation of
the RCS Reference Model Architecture Using an Architectural Description
Language," Lecture Notes in Computer Science EUROCAST ’99, Volume1798,
April 2000. pdf version
Balakirsky, S. and Lacaze, A., "World Modeling and Behavior Generation
for Autonomous Ground Vehicles," Proceedings of the 2000 IEEE International
Conference on Robotics and Automation, April 2000. pdf
version
Neal, R. and Messina, E. R., Editors, Proceedings of the First Part
Correct Workshop, Gaithersburg, MD, April 2000.
Horst, John, "Architecture, Design Methodology, and Component-Based
Tools for a Real-Time Inspection System," Proceedings of the 3rd IEEE
International Symposium on Object-oriented Real-time distributed Computing
(ISORC 2000), Newport Beach, California, March 15 - 17, 2000. pdf
version
Messina, E., Horst, J., Kramer, T., Huang, H., Tsai, T., and Amatucci,
E., "A Knowledge-Based Inspection Workstation," Proceedings of the 1999
IEEE International Conference on Information, Intelligence, and Systems,
November 1999. pdf
version
Messina, E., Horst, J., Kramer, T., Huang, H., Michaloski, J., "Component
Specifications for Robotics Integration," Autonomous Robots Journal,
Vol. 6, No. 3, June 1999. pdf
version
Albus, J., "The Engineering of Mind," Information Sciences,
Vol. 117, No. 1-2, 1999. pdf
version
Projects Associated with this Program
Performance Metrics
Knowledge Engineering
Architectures and Tools
Learning and Adaptive Control
DARPA Mobile Autonomous Robot Software
Contact
Ms. Elena
Messina
NIST
Intelligent Systems Division
100 Bureau Drive, Mail Stop 8230
Gaithersburg, MD 20899-8230
Phone: 301-975-3510
Fax: 301-990-9688
E-mail: elena.messina@nist.gov
isd-webmaster@cme.nist.gov
Date created: 1/25/2001
Last updated: February 22, 2002
|
