Ontology-based knowledge model for human-robot interactive services

Abstract

Recently, increasing interest in service robots with artificial intelligence has triggered several studies on developing service robots as human assistants. To perform automated tasks, service robots are not only required to recognize various attributes of the service environment, but they must also determine, which tasks and behaviors to perform, according to their internal system specifications and those of the individual situation. To perform tasks in such a generalized manner efficiently, the service robot must abstractly understand the recognition data obtained from the external environment and plan its tasks by combining the data and existing knowledge. Thus, an abstract and integrated knowledge management of low-level external recognition data and symbolic-level knowledge is indispensable. This study proposes a knowledge model for an integrated robot framework, which is used to provide human-robot interactive services. Our knowledge model is based on an ontology that contains general knowledge which provides clear definitions of common concepts and domain knowledge which provides specific concepts required to understand the information about agents(users and robots), and the environments about human-robot interactive services. An exemplary experiment is given in the context of a social robot service, which shows the usability of our knowledge model.