Coordinating Agents in OO

Abstract

In this paper we introduce an object-oriented coordination language for multi-agents systems. The beliefs and reasoning capabilities of an agent are specified in terms of a corresponding abstract data type. Agents interact via an extension of the usual object-oriented message passing mechanism. This extension provides the autonomy that is required of agents but which objects in most object-oriented languages do not have. It consists of an explicit answer statement by means of which an agent can specify that it is willing to accept some specified messages. For our coordination language we also present a formal method for proving correctness. The method extends and generalizes existing assertional proof methods for object-oriented languages.