B.3 NIAD&R: the Distributed AI & Robotics Group

B.3 NIAD&R: the Distributed AI & Robotics Group

We have selected three main research lines in the Autonomous Agents field that we believe can be important also in the near future:

• Flexible Tools and trustful Frameworks for agents interaction. We believe that sophistication of autonomous and pro-active agents can only be relevant when people understand and trust the environment in which their procurers (intelligent software agents) perform their activities on their behalf. This is why our research aims at the progressive establishment of largely accepted Electronic Institutions providing, on one hand, all the needed capabilities for fair, however personalized negotiation protocols and agents' coordination and, on the other hand, the capability "to look over the shoulder" of what is happening during the agents' electronic encounters. We are pursuing this research line mainly in the domain of Virtual Enterprises life cycle.
• Advanced features for Autonomous Agents. A set of essential, intrinsic features are needed for computational agents to show intelligent (sometimes human-like) behaviour. The quest for real learning and flexible adaptation is a difficult one. Due to our current interests, we have elected as our research objective in what this topic is concerned, enhancing agents' on-line (therefore not using traditional off-line examples-based) learning policies. Also in the same line, we are now exploring how "emotion-based" architectures can lead agents to display decision-making capabilities that are not completely under the utility-based function paradigm.
• Multi-Agent coordination and cooperation in dynamic environments. Our aim is to derive methods and protocols for coordination and cooperation of agent teams that perform complex tasks in dynamic, inaccessible, non-deterministic multi-agent scenarios.

We are currently trying to apply agent-based techniques to Virtual Enterprises formation as well as to Electronic Markets. Electronic Markets for Electrical Energy is one of our applications in progress. Automatic agent-mediated e-Brokering for the Insurance domain, is another problem we have modeled, developed and proposed an interesting (feasible) solution to.

Distributed Resources management in the context of Civil Construction companies and non-monotonic decision support systems for Projects location were also problems we have recently dealt with that have produced (in 2001) available prototypes.

We may summarize the scope of our research efforts by explicitly emphasising the following topics: Adaptive, multi-attribute agents negotiation; emotion-like agents architectures; learning capabilities and decision-making in Multi-Agent Systems; Agent-based teams coordination.

We also have been, for the last years, present in some of the major international agent related conferences Programme Committees (AAMAS- Autonomous Agents and Multiagent Systems, Autonomous Agents, CIA-Cooperative Information Agents Workshop, IEEE International Conference on WebIntelligence) as well as ICMAS, MAAMAW, IEEE International Conference on Concurrent Engineering.

We have been co-founders of the European Network of Excellence AgentLink, now AgentLink II. We have been active, since the beginning, in both the AMEC(Agent-mediated Electronic Commerce) and ALAD (Adaptive and Learning Agents) Special Interest Groups, organising special sessions and presenting our current research. We also have contributed to the book already published by AMEC SIG.

We have submitted project proposals together with international research teams and we still are collaborating in new proposals to be submitted to the 6th European Framework Programme. We have also successfully participated in the major robotic soccer (RoboCup) competitions. In previous years, we have won one RoboCup championship (simulation league Melbourne 2000), two European championships (simulation league Amsterdam 2000 and Paderborn 2001) and one RoboCup world coach competition (Fukuoka, 2002). We are currently supervising students both coming from foreign Universities (Brasil) and staying at foreign companies (Intellix/Denmark, Infopulse BV in Holland). We have agreements under the Sokrates European Funding Programme enabling both students as well as professors exchanges with University of Trier (Germany), École des Mines de Saint Étienne (France) and Haute École de Liége (Belgium).

The group's coordinator belongs to the Editorial Board of the Autonomous Agents and Multiagent Systems International Journal (Kluwer) and to the European Board of IOS and Ohmsha Ltd. "Frontiers in AI and its Applications" books sub-series.

International cooperation

• AgentLink II: AMEC - Agent-Mediated Electronic Commerce SIG.
• AgentLink II: ALAD - Adaptive Learning Agents SIG.
• Imperial College- University of London, Prof. Abe Mamdani.
• Université de Technologie de Compiègne,France, Prof. Jean Paul Barthès.
• University of York, United Kingdom, Dr. Dimitar Kazakov, Dr. Daniel Kudenko.
• Achmea, The Netherlands, Dr. Virginia Dignum.
• Intellix, S/A, Denmark.
• DFKI -- German Artificial Intelligence Research Center, Saarbrücken, Dr. Klaus Fischer, Dr. Mathias Klush.
• École National des Mines de Saint-Étienne, France, Dr. Olivier Boissier.
• European Coordinating Committee for Artificial Intelligence, Dr.Rob Milner.
• Infopulse Electronic Commerce B.V., Holland.
• International Institute for Artificial Intelligence, Spain, Dr. Carles Sierra.
• Universidade Católica de Curitiba, Brasil, Dr. Flávio Bortolosi, Dr. Marcus Shmeil.
• University of Bath, United Kingdom, Prof. Julian Padget.
• University of Trier, Germany, Dr. Norbert Kuhn.
• University of Utrecht, The Netherlands, Dr. Frank Dignum.
• Xerox Research Centre Europe, France, Dr. Jean-Marc Andreolli.
• Haute École de la Province de Liége, Belgium.
• City College, University of London, Dr. Eduardo Alonso.

Team

B.3 NIAD&R: the Distributed AI & Robotics Group