Plan Representations for Distributed Planning and Execution

Abstract

Abstract : This report results from a contract tasking University of Edinburgh as follows: The final project report describes work completed towards a new framework for distributed multi-agent planning and plan execution. The first step focuses on the plan representation that needs to be sufficiently rich to allow the sharing of plans between humans, software systems, and robotic entities. The basis for this representation is the Core Plan Representation developed in the mid 90s. This representation and its roots in Artificial Planning research are examined in section 3. The focus of this representation is the concept of a plan. Plans are executed by agents, and one of the dominant models of agency in agent research describes three mental attitudes of an agent: beliefs, desires and intentions. In section 4 we shall examine these concepts attempt to merge them with the plan representation. The combined ontology will then be further refined in section 5, introducing concepts that will be required for a sharable plan representation intended for distributed multi-agent planning and plan execution. The report then goes on to describe the software support for this representation, which has been developed as an extension to MediaWiki. The next part of this report defines a number of features that can be used to characterize planning domains, namely domain types, relation fluency, inconsistent effects and reversible actions. These features can be used to provide additional information about the operators defined in a strips-like planning domain. Furthermore, the values of these features may be extracted automatically; efficient algorithms for this are described in this report. Alternatively, where these values are specified explicitly by the domain author, the extracted values can be used to validate the consistency of the domain, thus supporting the knowledge engineering process.