STRPLAN: A Distributed Planner for Object-Centred Application Domains

作者:Rafael Berlanga Llavori

摘要

This paper deals with the problem of constructing valid planning models as well as efficient planning techniques for multi-agent domain applications. For these purposes, we first introduce an object-oriented formalism that allows for a fully-structured description of all the elements of a planning theory. This formalism is intended to facilitate the acquisition, maintenance and re-use of the domain knowledge. The proposed formalism also serves as the basis for identifying useful structural domain properties. Afterwards, we introduce a new efficient multi-agent distributed planner, STRPLAN, specially suited for these models. STRPLAN relies on the concept of domain region, which takes into consideration the structural domain properties as well as the interference of actions. As a result, global plans can be decomposed into local linear plans as a function of these regions. Preliminary empirical results have shown that the presented approach is very efficient for highly-structured multi-agent domains. In this way, the proposed methodology provides a practical framework to treat large multi-agent domain applications.

论文关键词:planning theory, knowledge modelling and object orientation

论文评审过程:

论文官网地址:https://doi.org/10.1023/A:1008336121961