A semantically-based component selection mechanism for intelligent service robots

Abstract

Intelligent Service Robots sense environment, recognize troubled situations, determine how to solve the problems and perform relevant behaviors to overcome the situation. We cannot pre-code all robot behaviors to prepare for all situations since real environment around ISRs is dynamic and unpredictable. As a way to successfully and continuously provide services, Intelligent Service Robots system dynamically reconfigures their software with components which have functionalities to solve troubled situations. A component selection mechanism is an essential function to support dynamic reconfiguration. We follow a semantically-based approach for component selection in Intelligent Service Robots domain since the existing non-semantically-based component selection approaches have limitation. However component selection for Intelligent Service Robots is different from the existing semantically-based approaches in that situational information is more important rather than user requirements. Therefore, this thesis proposes a semantically-based component selection mechanism that utilizes situational information using ontology for Intelligent Service Robots. In the component selection, we adjust a relaxation technique to get relevant components more. In addition, to rank the components which are retrieved through relaxation, we measure semantic distances in ontology trees.