Strategy selection-based meta-level reasoning for multi-agent problem-solving

Abstract

Researchers have developed various techniques to address MAS problem- solving activities,i.e., agent organization construction, plan generation, task allocation, plan integration, and plan execution. An agent.s respective problem solving and coordination techniques must be properly understood before they can be included into any other software system. 'Strategies' describe the techniques by which agents perform their individual decision-making processes and coordinate those processes with other agents. This chapter describes current work in characterizing agent operations, specifically, the representation of strategies in terms of roles and interactions as well as a trade-off evaluation mechanism for deciding which strategy is most appropriate for a given situation. On-line evaluation and selection of strategies will allow agents to tailor their behavior to given environment situations and thus, offer increase flexibility and adaptability of response © 2001 Springer Berlin Heidelberg.