Keywords multi-agent systems, human computer cooperative work
Start Date: 01-NOV-90 / Duration: 30 months
[ contact / participants ]
The IMAGINE project aims to provide a sophisticated and comprehensive environment upon which a variety of complex multi-agent systems can be built. In particular the environment has to support applications in the area of Human Computer Cooperative Work, which require a high level of cooperation among collections of multidimensionally distributed, heterogeneous autonomous agents, including humans.
The major results achieved in the project are:
The analysis of various application scenarios has led to a conceptual framework required to build multi-agent systems in different domains. The basic concepts are: a communication framework for the underlying information flow between agents, an agent formalism providing a description level for individual agents, and a cooperation framework providing cooperation primitives that can be composed into cooperation methods to support a wide variety of cooperation styles.
MAIL: a Multi-Agent Interaction Language
Based upon the concept formalisation the multi-agent language MAIL has been developed, defining and implementing the fundamental "basic loop" of an agent allowing for the definition of a wide variety of agents, their respective communication capabilities and cooperative behavior.
APRIL: an Agent Process Interaction Language
A process-oriented language, APRIL, with dedicated facilities (distribution, symbolic data structures and pattern matching, uniform message passing) tailored for the implementation of higher multi-agent formalisms has been provided. This language constitutes both the underlying computational model for the final implementation of MAIL and an independent distributed language.
MECCA: a Multi-Agent Environment for Constructing Cooperative Applications
As a result of a conceptual prototyping approach an integrated environment has been developed comprising a first version of MAIL and the tool architecture. It is implemented in IC-Prolog II, a distributed logic language platform which integrates a multi-threaded version of IC-Prolog and Parlog. IC-Prolog II has been provided within IMAGINE as a powerful prototyping platform.
Application prototypes have been developed based on the MECCA/MAIL and APRIL platform in the following application areas: urban traffic control (MECCA/UTS), appointment management (MECCA/AM), airport catering and workflow management
Both the platforms and the applications are used by the partners for further development and exploitation at the tool and application levels.
SIEMENS is developing internal multi-agent systems based on IMAGINE results, by incorporating human-human co-operation facilities. An urban traffic control architecture environment is being built as well as other environmental control, emergency management and distributed computing applications.
STERIA is pursuing the development of multi-agent distributed industrial applications.
INTRASOFT uses the experience gained in IMAGINE in distributed scheduling and planning and in tools development through project 6310 MMTCA.
Two small spin-off companies have been created: SYSTEMS ENGINEERING Ltd. and Human Resources Engineering which address the multi-agent and human co-operative work respectively.
Mr Hans Haugeneder
Otto Hahn Ring 6
D - 81730 MÜNCHEN
tel: + 49/ 89-636-43698
fax: + 49/ 89-636-42284
SIEMENS AG - D - C
INTRASOFT SA - GR - P
STERIA - F - P
ROKE MANOR RESEARCH LTD - UK - P
IMPERIAL COLLEGE OF SCIENCE,
TECHNOLOGY & MEDICINE - UK - A
UNIVERSITY OF KEELE - UK - A
UNIVERSITEIT VAN AMSTERDAM - NL - A
IMAGINE - 5362, December 1993
please address enquiries to the ESPRIT Information Desk
html version of synopsis by Nick Cook