Introduction to the Industry and Application Track

AAMAS
Autonomous agents and multi-agent systems are no doubt interesting paradigms for software modeling and development and raise much hope in the community for the last twenty years:

  • Autonomous agents are computer programs capable of deciding between different methods of achieving their programmed goals. This approach is in contrast with procedural and predefined behavior promoted by the traditional methods because autonomous agents can cope with dynamic changes of their environment,
  • Multi-agent systems are environments where different entities possibly developed independently, can discover each other, coordinate their behaviors and achieve global goals as well as individual ones. Therefore multi-agent technology seems particularly applicable to large and scalable systems discovering each other and coordinating their efforts towards larger goals realizing self-organizing and –optimizing behavior.

By combining autonomy of agents into environments where agents can reason, discover and embrace evolution, the AAMAS community has developed a very promising step forward in computer science. However, the AAMAS community has long recognized that the significance of agent abstraction, techniques, tools and technologies developed by the research community will only be successful if they have practical applications.

The AAMAS Industry and Applications Track Philosophy
The goal of the AAMAS Industry and Applications Track is to lead agent technology towards applications, to define problem spaces, agents use cases and domains of excellence where agents outperform other techniques to establish an agent stronghold and foster large-scale deployment of this technology. To reflect this mission, the Industry and Applications Track introduced, in addition to the classic reviewing criteria (originality, readability, technical quality and soundness), three new ones expressing the quintessence of industry applications. These criteria are systematically used to screen projects in nearly all industries:

  • Significance of problem domain, user needs, use cases, applicability – in terms of business size or cost of problem to solve,
  • Market potential / business case deployment and adoption: measuring the strength of the business case ranging from none to a full business description, including business evaluations and differentiation strategy,
  • Deployment and adoption: measuring the size of deployment covering – none, lab, field trial >20 users, medium size deployment > 1000 users, and large scale deployment > 10000 users.

These criteria define quite precisely the philosophy of this Industry and Applications Track. They are tough but reflect well the difficulty of deploying new technologies, such as agent technologies, to the industry. To win its place in the industry agents need either:

  • To outperform existing methods in a very convincing manner by making a major breakthrough that justifies replacing trusted existing technologies with a very new but unproven approach, or
  • To integrate and leverage proven industry algorithms (e.g. dynamic workflows, scheduling, coordination, constraint programming, market places…) with agents providing optimum solutions linking autonomy of agents to well known and industry trusted problem solving methods, or
  • To target and model very specific use cases and problems particularly suited to the agent approach, and to link these models with real world applications.

Therefore this AAMAS Industry and Applications Track pays a lot of attention to metrics, empirical measurements, simulations and comparison to established methods in real-size problems.

Papers
The AAMAS Industry and Applications Track 2008 is proud to present 12 long papers and six short ones. These papers were selected out of 30 submissions through the independent Industry and Applications Track reviewing process using the above mentioned criteria. Each paper was evaluated by three independent reviewers.

The trends of this track are:

  • Business process management: description of the integration of agents into business processes as well as in Workflow engines. Applications on multimedia distribution, manufacturing and healthcare workflows,
  • Procurement and contracting: use of electronic contracting to manage optimally supply chains organizing physical entities, schedules, locations, bottlenecks, transports. This is applied to the field of aerospace applications, insurance, and SLA in software engineering,
  • Aerospace applications: Agents approaches are used for aero engine overhaul, optimization of UVA mission control and of air traffic control,
  • Manufacturing and logistics introduces auction-based allocations of load in transportation logistic as well the optimization of stock control in virtual warehouses. Another case study optimizes a supply chain in the steel production process,
  • Energy management, optimization of power management in data centers, and optimization of pipeline operations in the oil industry,
  • Security and surveillance, one study deployed agent technology to detect the intrusion in a system, while another one leveraged game theory to optimize the security at LAX airport.

Several of these papers represent years of efforts for the deployment of these applications leading from initial ideas to models, proof of concept sometimes through several publications, including some in the AAMAS main track, before the work could eventually find its niche where these applications compete successfully with established non-agent technologies.

We commend the authors of these papers for their pioneering work and the risks they take to impose agent technology as an industry solution. We look forward to the presentation of the technical program and the discussions that will ensue.

Keynote Presentation
Christian Dannegger from Whitestein Technologies describes a machine control system applied to control a modular soldering machine. Agents are attached to each machine module as well as to each of the boards to be soldered. This agent system optimizes: the module's operation, the soldering parameters for each board, as well as the capacity production while providing increased flexibility, adaptability and maintainability. It is planned for worldwide distribution from mid 2008.

Acknowledgements
The co-chairs wish to thank the AAMAS organizers for their support for the Industry and Applications Track and the Industry and Applications Track Program Committee for their outstanding work in reviewing the papers according to the philosophy of this track and providing insightful comments to authors.

Special thanks go to Jörg Müller for his constant support, Lin Padgham, David Parkes and other chairs for arranging the conference tracks and accommodating for the Industry and Applications Track, and finally, to Franziska Klügl for all the work linked to the final edition of these proceedings.

Michael Berger, Bernard Burg, Satoshi Nishiyama
May 2008

    
 

 

  
end of page