An Ontology-based Cooperative Environment for Real-world Agents

摘要

In this paper, we propose a framework for communication and cooperation among heterogeneous real-world agents based on shared ontologies. Ontologies give a background of knowledge to share among agents, that is, systems of concepts are defined which are used to communicate to each other. We present three different ontolagies, namely, ontologies for object, space, and action. Based on these ontologies, we then show mediation mechanism to generate action sequences for agents from given tasks. Our framework is appropriate for cooperation among heterogeneous real-world agents because of (1) adaptiveness for environments and agents, (2) implicit representation of cooperation, and (3) integration of information and real-world agents. We realized and verified our mediation mechanism by applying task planning for mobile robots and computer-controlled instruments.