Automatic ontology creation, aiming to develop ontology without or with minimal human intervention, is needed for robots that work in dynamic environments. This is particularly required for service (or domestic) robots that work in unstructured and dynamic domestic environments, as robots and their human users share the same space. Most current works adopt learning to build the ontology in terms of defining concepts and relations of concepts, from various data and information resources. Given the partial or incomplete information often observed by robots in domestic environments, identifying useful data and information and extracting concepts and relations is challenging. In addition, more types of relations which do not appear in current approaches for service robots such as “HasA” and “MadeOf”, as well as semantic knowledge, are needed for domestic robots to cope with uncertainties during human–robot interaction. This research has developed a framework, called Data-Information Retrieval based Automated Ontology Framework (DIRAOF), that is able to identify the useful data and information, to define concepts according to the data and information collected, to define the “is-a” relation, “HasA” relation and “MadeOf” relation, which are not seen in other works, to evaluate the concepts and relations. The framework is also able to develop semantic knowledge in terms of location and time for robots, and a recency and frequency based algorithm that uses the semantic knowledge to locate objects in domestic environments. Experimental results show that the robots are able to create ontology components with correctness of 86.5% from 200 random object names and to associate semantic knowledge of physical objects by presenting tracking instances. The DIRAOF framework is able to build up an ontology for domestic robots without human intervention.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:738079 |
| Date | January 2017 |
| Creators | Kanjaruek, Saranya |
| Publisher | University of Bedfordshire |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://hdl.handle.net/10547/622481 |
Page generated in 0.0024 seconds