Semantic Representations of Agent Plans and Planning Problem Domains

Integrating knowledge representation approaches with agent programming and automated planning is still an open research challenge. To explore the combination of those techniques, we present a semantic model of planning domains that can be converted to both agent programming plans as well as planning problem definitions. Our approach allows the representation of agent plans using ontologies, enabling the integration of different formalisms since the knowledge in the ontology can be reused by several systems and applications. Ontologies enable the use of semantic reasoning in planning and agent systems, and such semantic web technologies are significant current research trends. This paper presents our planning ontology, exemplify its use with an instantiation, and shows how to translate between ontology, agent code, and planning specifications. Algorithms to convert between these formalisms are shown, and we also discuss future directions towards the integration of semantic representation, automated planning, and agent programming.