Automatically Learning Robot Domain Ontology from Collective Knowledge for Home Service Robots

Today, for enabling intelligent decision and high accuracy of recognition in service robots, many researchers supplement robot's knowledge model using the additional knowledge. However, the construction of them knowledge requiring much effort and domain experts fully depends on man power by few people. Thus, this paper proposes a fully automated process of acquiring domain knowledge and representing them to efficient and semantically abundant structure. Thus, we investigate the characteristics of OMICS as preceding case study for collective knowledge in robot domain, and describe the automated process of conversion of such collective knowledge to robot domain ontology. Also, we suggest dynamic semantic distribution method to solve appropriate generalization of relation problem. Finally, we evaluate the efficiency and semantic of our structure for the ontology compared to other knowledge bases for robots.