Conversation system for robot partners based on informationally structured space

This paper discusses a conversation system for robot partners based on informationally structured space using sensor networks. The conversation system is composed of (A) information support mode, (B) daily conversation mode, and (C) scenario conversation mode. The sensor network can extract the states on the person and environment through the informationally structured space, and the information support provides useful information in the daily life with a person. We apply a spiking neural network to extract environmental information. The proposed conversation system uses the information referred from the informationally structured space, and therefore, the same control architecture is applied to various robot partners. The experimental results show the effectiveness of the proposed method.