Requirements for Achieving Software Agents Autonomy and Defining their Responsibility

This paper addresses the problem of implementing agent-based software systems with respect to agent framework fundamental concepts such as autonomy and interaction without specifying any particular agent internal architecture. The autonomy and interaction axioms imply that a deployment environment has to be defined in order to achieve interaction among agents. This deployment environment may also encode environmental rules and norms of the agent society. The responsibility of an agent is then defined as being in adequacy with its environmental rules. Finally, a formal deployment environment, named MIC*, is presented with a simple application showing how interaction protocols are guaranteed by the deployment environment, which protects agents from non-conform actions and preserve their autonomy.

[1]  Yves Demazeau,et al.  A Social Reasoning Mechanism Based On Dependence Networks , 1997, ECAI.

[2]  Peter Norvig,et al.  Artificial Intelligence: A Modern Approach , 1995 .

[3]  J. Michael Spivey,et al.  The Z notation - a reference manual , 1992, Prentice Hall International Series in Computer Science.

[4]  Michael Luck,et al.  A Formal View of Social Dependence Networks , 1995, DAI.

[5]  Luc Steels,et al.  When are robots intelligent autonomous agents? , 1995, Robotics Auton. Syst..

[6]  Fabien Michel,et al.  MIC: An Agent Formal Environment , 2003 .

[7]  Michael Luck,et al.  A Formal Framework for Agency and Autonomy , 1995, ICMAS.

[8]  Cristiano Castelfranchi,et al.  Guarantees for Autonomy in Cognitive Agent Architecture , 1995, ECAI Workshop on Agent Theories, Architectures, and Languages.

[9]  Jacques Ferber,et al.  Les Systèmes multi-agents: vers une intelligence collective , 1995 .

[10]  Michael R. Genesereth,et al.  Software agents , 1994, CACM.