Exploiting agent diagnosis for plan repair in MAS

ABSTRACT

In this paper we formalize a local strategy for plan repair in a Multi Agent Plan, where agents exhibit a collaborative behavior to reach a common global goal in a partially observable environment. The recovery strategy is based on a plan failure analysis where two main results are inferred: the agent diagnosis (which explains the action failure in terms of faults in agent functionalities) and the set of missing goals (i.e. services that the faulty agent can no longer provide to the other agents). Relying on the agent diagnosis, a faulty agent activates a local re-planner to restore the healthy status in its functionalities and achieve the set of missing goals.

REFERENCES

• 1: Jeffrey S. Cox , Edmund H. Durfee , Thomas Bartold, A distributed framework for solving the Multiagent Plan Coordination Problem, Proceedings of the fourth international joint conference on Autonomous agents and multiagent systems, July 25-29, 2005, The Netherlands [doi>10.1145/1082473.1082598]
• 2: Keith Decker , Jinjiang Li, Coordinating Mutually Exclusive Resources using GPGP, Autonomous Agents and Multi-Agent Systems, v.3 n.2, p.133-157, June 2000 [doi>10.1023/A:1010074611407]
• 3: Bryan Horling , Brett Benyo , Victor Lesser, Using self-diagnosis to adapt organizational structures, Proceedings of the fifth international conference on Autonomous agents, p.529-536, May 2001, Montreal, Quebec, Canada [doi>10.1145/375735.376436]
• 4: R. Micalizio and P. Torasso. Recovery from plan failures in partially observable environments. In Research and Development in Intelligent Systems XXVII, pages 321--334, 2007.
• 5: Roberto Micalizio , Pietro Torasso, Monitoring the Execution of a Multi-Agent Plan: Dealing with Partial Observability, Proceeding of the 2008 conference on ECAI 2008: 18th European Conference on Artificial Intelligence, p.408-412, June 27, 2008
• 6: R. van der Krogt and de Weerdt. Plan repair as an extension of planning. In Proc. of ICAPS'05, pages 284--259, 2005.

INDEX TERMS

Primary Classification:
  I. Computing Methodologies
  I.2 ARTIFICIAL INTELLIGENCE
        I.2.11 Distributed Artificial Intelligence

Additional Classification:
  I. Computing Methodologies
  I.2 ARTIFICIAL INTELLIGENCE
      I.2.8 Problem Solving, Control Methods, and Search
          Subjects: Plan execution, formation, and generation

Keywords:
agent reasoning, multi-agent plan diagnosis, multi-agent plan recovery