Object Services and Consulting, Inc.

  Agility: Agent -Ility Architecture

FINAL REPORT

June 17, 2002

Principal Investigator: Craig Thompson
 Team:  Tom Bannon, Steve Ford, Paul Pazandak
Subcontractors:  none

DARPA Control of Agent-based Systems (CoABS)
DARPA Program Manager:  Dr. James Hendler, LCDR Dylan Schmorrow
COTR:  Wayne Bosco, AFRL

Contract No. F30602-98-C-0159  AO J357
Contract Duration:  18 June 1998 - 17 June 2002

This research is sponsored by the Defense Advanced Research Projects Agency and managed by the U.S. Air Force Research Laboratory under contract F30602-98-C-0159. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the Defense Advanced Research Projects Agency, U.S. Air Force Research Laboratory, or the United States Government.


Executive Summary

We provide four top-level project summaries: In addition, we provide specific descriptions and user manuals for the three technologies we developed:  eGents, AgentGram, and WebTrader.

Quad Chart Overview


Problem

Today's agent systems are monolithic, centralized, and do not provide a clear migration path for integration with mainstream technologies (e.g., object and web technologies).  As a consequence, though agent technology is identified as a promising high impact DoD software technology, it is not significantly impacting DoD, software technology or the mass market.

Objective

The objective of the Agility project is to develop an open agent grid architecture populated with scalable, deployable, industrial strength agent grid components, targeting the theme "agents for the masses."

Approach

Our overall technical approach has been to deconstruct agent systems into components, then populate an open agent grid architecture with scalable light-weight agent grid components that are engineered to piggyback on existing and emerging standards (e.g., distributed objects, email, web, search engines, XML, Java, Jini).

Specific objectives are to develop:

These components should operate standalone and/or interoperate with the CoABS grid.

Accomplishments

Architecture

In 1998 we presented Strawman Agent Architecture to DARPA CoABS, ATAIS, and ALP, then submitted it to OMG Agent SIG and to FIPA.  Parts of the presentation are now incorporated in the OMG Agent SIG Green Paper on Agent Technology, which we co-authored.

In 1999, our paper Characterizing the Agent Grid was submitted to and accepted for Jeff Bradshaw's book Handbook of Agent Systems.  Sections of the paper appear in the architectural sections of the CoABS Grid Vision document (available on the GITI CoABS website, password protected).

Prototypes

To demonstrate our component-based approach, we are developed prototypes of three agent grid components (eGents, AgentGram, and WebTrader), and made the following technical progress:

Technology Transition

Papers and Workshops

In addition to our architecture presentation and grid paper mentioned above, early versions of the three prototypes were described in a workshop paper: Also, we were/are on the program committees for:

Software and the Grid

All three prototypes were extended to operate over the over the CoABS 24x7 Grid and were available for a year on the grid (see the grid archives and individual descriptions of eGents, AgentGram, and WebTrader).

TIEs

We participated in four TIE efforts covering DoD domains:

Patents

We completed two patent applications:

Related DARPA Programs and Other Contracts

Frank Manola completed the ALP-CoABS Integration Final Report in July 1999.   Program managers Jim Hendler and Todd Carrico indicated that we did a good job and both felt more integration work might occur at a later date but neither was ready to fund more work at that time on CoABS-ALP TIEs.

Work begun on eGents under CoABS has led to two additional contracts:

Standards

We influenced agent standards:

Commercialization

We are working on productizing AgentGram via the spinout LingoLogic.com.

Future Directions

Some of the more important future directions are:

Additional next steps for eGents, WebTrader, and AgentGram are described in the Final Review .ppt presentation.

Conclusion

We started with a reasonable thesis and were able to demonstrate feasibility of the approach of deconstructing agent systems into components and rebuilding industustrial strength components, piggybacking their implementations on already pervasive technology.  This does appear to be a viable route for getting agent technology into the mass market in a way that provides a migration path.