Executive Summary:

**Papers to be published in edited collection. Much work in multi-agent systems focuses on coordinating the activities of agents so that the end result approximates the solutions possible if one were to centralize the activities being carried out by the agents. The approach taken to coordination, or even whether coordination is necessary, is often dependent on certain application features. For instance, in supply chain management agents may need to reason about temporal constraints whereas in RoboCup a more reactive coordination approach may be appropriate. In this workshop we will examine the landscape of MAS applications/problem domains and attempt to characterize, classify, and differentiate different problem spaces to understand how problem domains and coordination/control techniques relate.

Workshop Schedule:

The Workshop is on Monday the 15th. 8:30 - 9:00 Introductions and opening remarks by Tom Wagner. 9:00 - 9:40 Invited talk: Scaling Up Agent Coordination Strategies by Edmund H. Durfee. 9:40 - 10:10 Invited talk: A complex Systems Perspective on Collaborative Design by Mark Klein, Peyman Faratin, Hiroki Sayama and Yaneer Bar-Yam. 10:10 - 10:30 Short research overview presentations. Reconciling Objective and subjective Coordination in Multiagent Systems, by Alessandro Ricci. A Multi-Agent Approach for Tactical Aircraft Simulation, by Irene Degirmenciyan-Cartault, Frederic Marc and Amal El Fallah-Seghrouchni. Supporting Life Cycle Coordination in Open Agent Systems, by F.M.T. Brazier, D.G.A. Mobach, B.J. Overeinder and N.J.E. Wijngaards. 10:30 - 11:00 Coffee break. 11:00 - 11:25 Applying coordination mechanisms for Dependency Relationships under Various Environments by Wei Chen and Keith Decker. 11:25 - 11:50 Importance and Properties of Roles in MAS Organization: A review of methodologies and system, by Ioannis Partsakoulakis and George Vouros. 11:50 - 12:30 Panel I - Comparing technologies and techniques. Panelists: Durfee, Klien, Vouros, Decker. 12:30 - 1:30 - Lunch. 1:30 - 1:55 Analyzing Factors Influencing Decision making Frameworks and Impacting Agent and Multi-Agent System Performance by K. Suzanne Barber and Matthew T. MacMahon. 1:55 - 2:20 Characterization and Evaluation of Distributed Just-In-Time Production and Distribution by Paul Davidsson and Fredrik Wernstedt. 2:20 - 2:45 Coordination Challenges for Autonomous Spacecraft, by Bradley J. Clement and Anthony C. Barrett. 2:45 - 3:30 Panel II - Comparing problem spaces. Panelists: Joerg Mueller, Clement, Barber, Wagner. 3:30 - 4:00 Coffee break. 4:00 - 4:20 Short research overview presentations. Dynamic Agent Replication Groups, by Alan Fedoruk and Ralph Deters. An Experiment In Automatic Adaptor Generation by means of Evolution, by Werner VanBelle and Theo D'Hondt. Multi-Agents Systems for Aeronotic Simulation, by Okba Kazar, Zaidi Sahnoun and Louis Frecon. 4:20 - 4:45 An Evolutionary Framework for Large-scale Experimentation in Multi-agent Systems by Alex Babanov, Wolfgang Ketter and Maria Gini. 4:45 - 5:10 Multi-Agent System Interaction Protocols in a Dynamically Changing Environment, by Martin Purvis, Stephen Cranefield, Mariusz Nowostawski and Maryam Purvis. 5:10 - 5:30/x Closing group discussion: toward an application science -- next steps.

---

Workshop Program:

Panel I - Comparing technologies and problem spaces. Participants being determined. Panel II - Developing a science of application. Participants being determined. Presentations: Scaling Up Agent Coordination Strategies, by Edmund H. Durfee. A complex Systems Perspective on Collaborative Design, by Mark Klein, Peyman Faratin, Hiroki Sayama and Yaneer Bar-Yam. Importance and Properties of Roles in MAS Organization: A review of methodologies and system, by Ioannis Partsakoulakis and George Vouros. Analyzing Factors Influencing Decision making Frameworks and Impacting Agent and Multi-Agent System Performance, by K. Suzanne Barber and Matthew T. MacMahon. An Evolutionary Framework for Large-scale Experimentation in Multi-agent Systems, by Alex Babanov, Wolfgang Ketter and Maria Gini. Applying coordination mechanisms for Dependency Relationships under Various Environments, by Wei Chen and Keith Decker. Characterization and Evaluation of Distributed Just-In-Time Production and Distribution, by Paul Davidsson and Fredrik Wernstedt. Coordination Challenges for Autonomous Spacecraft, by Bradley J. Clement and Anthony C. Barrett. Multi-Agent System Interaction Protocols in a Dynamically Changing Environment, by Martin Purvis, Stephen Cranefield, Mariusz Nowostawski and Maryam Purvis. Short Presentations: Combining Global Specification and Autonomous Execution, by Alexander Wise, Barbara Staudt Lerner and Leon J. Osterweil. Reconciling Objective and subjective Coordination in Multiagent Systems, by Alessandro Ricci. A Multi-Agent Approach for Tactical Aircraft Simulation, by Irene Degirmenciyan-Cartault, Frederic Marc and Amal El Fallah-Seghrouchni. Supporting Life Cycle Coordination in Open Agent Systems, by F.M.T. Brazier, D.G.A. Mobach, B.J. Overeinder and N.J.E. Wijngaards. Dynamic Agent Replication Groups, by Alan Fedoruk and Ralph Deters. An Experiment In Automatic Adaptor Generation by means of Evolution, by Werner VanBelle and Theo D'Hondt. Multi-Agents Systems for Aeronotic Simulation, by Okba Kazar, Zaidi Sahnoun and Louis Frecon.

---

Topic:

One goal of multi-agent systems is to solve distributed problems in such a way as to approximate the solutions obtainable if all computation could be centralized. Much research in MAS is on exactly this problem -- how to make it so agents can act locally but achieve coherent global behavior. This class of research takes many forms and includes multi-agent coordination, negotiation, scheduling, planning, agent organization, and problem decomposition to name a few. From this view much of the control problem solving performed by agents in MAS is a form of distributed optimization. Properties of different problem spaces impact the degree to which this kind of distributed optimization is necessary and possible and accordingly the approach taken to coordinating the local decisions of individual agents. For instance, many successful strategies in robotic soccer do not require reasoning about deadlines or planning far downstream temporally -- this is in contrast to a supply chain problem or agent-driven logistics. In essence, the character of the distributed optimization problem and its attributes vary along many dimensions and these differences are driven by applications/problem spaces, different assumptions about problem decomposition, and the types and frequency of interactions between agents. These problems spaces directly impact agent coordination and other MAS technologies. In this workshop we will examine the landscape of MAS applications/problem domains and attempt to characterize, classify, and differentiate different problem spaces. We will also attempt to extract requirements from the domains and understand the implications of different domains to the optimization problems faced by the agents and the coordination technologies/approaches designed to perform the optimization. The long term objective is to move the agent community toward a scientific approach for mapping out a particular application domain and selecting a class of agent control technologies for dealing with the optimization problems present in the domain. Sample domains include but are not limited to: supply chain management, distributed business process coordination, ecommerce, coordination of teams of AUVs, multi-agent information gathering, sensor management, agent-based logistics, etc. Features we will likely examine and explore include: the need for coordination between the agents, the presence of deadlines or temporal constraints, the presence of interactions between agents (e.g., task, resource), required tempo/pace of decision making, level of uncertainty, sequential decision making aspects, the need for temporal synchronization, availability or scarcity of required (shared) resources, grain-size, complexity and scope of individual decision making, the degree to which non-local information is visible or obtainable, the uncertainty of the environment, dynamics, openness, predictability, etc.

Format:

Full day workshop. Paper presentations grouped according to topic. The workshop will emphasize group and panel discussion so presentations will be balanced with discussion time. Participants should submit either a paper for presentation or a research statement. Proposals for panels / discussion groups during the workshop are also welcome.

Suggested Paper Topics:

What properties of problem spaces make MAS coordination and optimization easier or harder? Under what circumstances is coordination unnecessary, irrelevant, or counter productive to achievement of globally coherent (or optimized) behavior? How should MAS design and organization be influenced by problem space properties and global performance objectives? What are the good problem spaces for MAS optimization? Are there problem spaces where some level of global competence or optimality can be reasonably ensured? Characterization of an application domain, e.g., RoboCup, supply chain management, coordination of unmanned attack vehicles, etc. Papers comparing and contrasting different domains are also encouraged. An analysis of key application domain features and how the features have influenced agent control technologies. For instance, identification of a set of supply chain problem space features and then identification of how dealing with these features changes/influences the design of the MAS and the coordination technologies needed by the individual agents. Exploration of how different domains impact coordination and how this flows through to the scheduling or planning technologies of individual agents. Papers that propose MAS coordination challenge problems. For instance, one might classify a set of domains and extract key features and then create a challenge problem that is representative of the application class.

Dates & Submission Details:

Submission of draft papers to organizers: April 29, 2002. Notification of acceptance: May 14, 2002. Submission of camera-ready papers to organizers: May 31, 2002. Workshop: July 15 or 16, 2002. Papers and statements should be submitted electronically in *PDF format* (formatted for US Letter) and sent to Dr. Tom Wagner via Ms. Judy Nyline at nyline_judy@htc.honeywell.com. If electronic submission is not possible, hardcopy should be directed to: Dr. Tom Wagner, 3660 Technology Drive (MN65-2600), Minneapolis, MN 55418. Questions may be directed to wagner_tom@htc.honeywell.com.

Organizers:

Tom Wagner - Honeywell Laboratories, wagner_tom@htc.honeywell.com George Vouros - University of the Aegean, georgev@aegean.gr Steven F. Smith - Carnegie Mellon University, sfs@cs.cmu.edu

Program Committee:

Marie desJardins - University of Maryland Baltimore County Suzanne Barber - University of Texas Austin Bernhard Bauer - Siemens Germany Brad Clement - Jet Propulsion Laboratory Keith Decker - University of Delaware Adele Howe - Colorado State University Yanis Labrou - PowerMarket, Inc. Michael Luck - University of Southampton Victor Lesser - University of Massachusetts Amherst Joerg Mueller - Siemens, Germany Martin Purvis - University of Otago Sandip Sen - University of Tulsa Peter Stone - AT&T Labs Katia Sycara - Carnegie Mellon Jose Vidal - University of South Carolina

Back to Tom Wagners homepage.