A proposal of communication design for continual human-agent interaction using natural utterances

In the fields of robotics and artificial intelligence, there have been gradually increasing the works dealing with the human-agent interactions, such as pet-robot developments for entertainment, humanoid robot applications, interactive teaching systems, and so on. The most of researchers engaged in these fields just have aimed at developing intelligent agents and referred to the unavoidable nature of human, i.e. we could implicitly adapt to the artificial systems. However, interacting continually with these agents makes human boring as time goes by, since humans could be gradually accustomed to the pre-designed behavior selection manners, which are embedded in them in advance. So as to actualized continual and active interactions between the agent and the user, it seems to be crucial that the both possess not only intrinsic behaviors but adaptation abilities and also affective systems. Based on the consideration, this paper proposes an interactive system design methodology by focusing on evaluation criteria for the adaptation mechanism of artificial agent.