出版社：Japan Society for Fuzzy Theory and Intelligent Informatics
摘要：This paper proposes an autonomous agent system that can recognize human intentions. Recently, the computers and the robots (agents) are used at any situation in the human life. However, the users and the systems cannot draw out each ability, because, in the present condition, only some I/O channels are allowed in the interaction between human and those systems. Furthermore, in expansion of the function, the users must remember many techniques. On the other hand, although humans do not use the some languages to communicate, they can still exchange their intentions. They can use various I/O channels combination that will recognize complex intention transmission. Therefore, even if they cannot understand partner's language, they can take some communications. We paid attention to the transmission of information by nonverbal information, such as gesture or body language used generally in human's communications. We applied the method of communications between humans to that between human and systems. And we aimed at organizing the autonomous agent that understands commands and situations by itself and acts, by artificial ontologies that are technological models of human intentions. In this system, each agent constructs artificial ontologies, and shares with other agents. As a result, this system can correspond to the different individual intention expression and environments. Artificial ontologies are constructed by Conceptual Fuzzy Sets, which is based on a bi-directional associative memory network. This system recognizes the human's action by a bi-directional associative memory. We show the effectiveness of the information sharing by artificial ontology in this agent system. The purpose of this research is to create human centered systems that can provide a natural interface in their interaction with people. This paper shows effectiveness to various human support robots.
