History of DAIAS- NIADSA (former DAI&R-NIAD&R)

"Distributed Artificial Intelligence and Agent-based Simulation" Group

 When, back in 1985, Eugenio Oliveira arrived from a stay of almost four years at the New University of Lisbon and about one year at IBM- International Education Center in La Hulpe (near Brussels, where he was invited as "guest academic"), he launched some small projects, both on Knowledge Based Systems and Planning Decision capabilities for Assembly Robotics. These seeds grew up and the results, although modest, were encouraging enough to permit diversification of efforts both on the application and theoretical side of AI.

On the Applications side, Expert Systems for diagnosis of Cardiac Arrhythmia and for Welding processes, Intelligent Tutoring Systems and Expert System Shells were among the first successful realizations. Also Symbolic Planners, taking into account physical and geometrical object constraints were developed and demonstrated.

Concerning the Theoretical aspect, we have started, at that time, studies on User Modelling, Natural Language (Portuguese) Interfaces and Reasoning with Uncertainty techniques.
 

 -  DAIAS-NIADSA (former DAI&R-NIAD&R) from 1989 to 2012

Since 1988, the main focus of attention of our Research Activities has been DISTRIBUTED ARTIFICIAL INTELLIGENCE. The main catalyst for this interest was, in the beginning, an important 5 years ESPRIT Project, ARCHON (Architecture for Cooperative Heterogeneous on-line Systems) in which we have been involved together with industry (Atlas Elektronik /Germany, Framentec/France, Iberdrola/ Spain, Volmac/Holland) and other Universities (U.London-QMWC, U.Amsterdam, NTU. Athens).

Again, our efforts have been made in two different fields: The Applications and the Theory. In which the Applications is concerned, we have developed a specific tool for the automatic generation of Cooperative Expert Systems. UPShell (University of Porto Shell) integrates different Expert Systems in a cooperating community permiting full cooperation through negotiation, based on relevant knowledge made available to each one of the Agents. UPShell has been applied to a scenario where 6 different Expert Systems have to cooperate in order to detect disturbances on a electricity distribution network, to preview black-out areas and to derive restoration plans to recover from failures.

Another application of the same principles of Cooperation, Distribution and Negotiation was in the Robotics area. A flexible Cell for Assembly including a manipulator and several sensors (laser ranging finder, Camera,...) is controlled by a Multi-Agent Systems which includes symbolic as well as geometrical planners , object recognition systems, object databases, low level manipulator controllers, etc.

We then extend the application of Multi-Agent Systems to different fields ranging from Geographical Information Systems and Simulation of Organizations interaction on the Market to Mobile Robotics.

On the Theoretical side, our major topics of interest at the time, were planing strategies for Agents, Conflict Resolution, Truth Maintenance in the Distributed environment and full cooperation of heterogeneous agents following both paradigms, reactive and cognitive. A mobile robot with arm was used to illustrate these concepts in a realistic scenario.

We developed Negotiation protocols leading to effective Agent coalitions. These policies were tested in the framework of the Resources Management (for Civil Construction Companies) domain.

Since 1998 our group was first an active node of the Network of Excellence AGENTLINK, and then of the "Agreement Technologies" European Network (COST Action). In this framework our main concern was the research of Flexible and adaptive protocols and tools for developing safe and trustworthy frameworks, seen as Electronic Institutions, Enterprise Networking and Virtual Enterprises formation.

In the last 5 years, the group, besides developing MAS architectures and methods for agents interoperability, has payed great attention to agent-based simulation principles and application of AI methods to Text Mining and Information Extraction from text.

As a consequence of very recent reshaping of LIACC Research Unit, the former group gave birth to a new separate group (HMIC- Human-Machine Intelligent Cooperation) and recentered on the area and topics of Distributed Artificial Intelligence and Agent-based Simulation.