Jirí Lazanský

Role of Qualitative Reasoning in a Multi-Agent System

The paper studies the possibility to use qualitative reasoning technology for improving behaviour of multi-agent systems. For this purpose, the twin-base communication model [Cao 97] is enriched to contain the state information (a tri-base model is defined) and it is combined with the idea of a super-agent. The super-agent exploring the qualitative reasoning methods acts as a global observer and state evaluator with limited reconfiguration rights. The impact and efficiency of the proposed approach are analysed.

Applications Model-Based Co-operative Agents in Production Planning

The tri-base acquaintance model of the agents behaviour is presented in this paper. This represents an extension of the twin-base model [Cao 97]. Based on practical experience, the new model tries to cope with parallel processing, precedence constraints, and sparse resources. The idea of substituting the interagent negotiation processes by the internal planning activity of the agents is stressed. This strategy is enabled by periodic revisions of the knowledge and data contained in the tri-base knowledge/data bases.

Does CIM Need AI

The parts of AI which seems to be of substantial importance for the CIM tasks are overviewed in this contribution. The stress is aimed mainly at impacts of Knowledge Engineering at the global CIM software architectures, at Distributed AI as a basis for system integration, at AI-centered view on Planning and Scheduling, at the role of Declarative Programming and Machine Learning in CIM tasks. The potential contributions of the stressed topics is discussed.

A Distributed System for CIM

This paper describes some basic issues and requirements for software structure of distributed CTM systems. Such systems have a wide applicability, especially in the context of geographically distributed manufacturing and business facilities. The paper describes also some introductory experiments on coupling and communication between the autonomous agents that act as stand-alone systems I and can cooperate.

Practical Applications of Planning Tasks

The AI literature very often investigates planning in an extremely abstract sense. The discussions on planning are rather sophisticated but usually the connection to the real-world tasks (as for instance industrial design problems) is quite obscure. The aim of this contribution is to refer to practical tasks where the AI planning methods may find its commercial value and where they can prove their strength.