Estimation of Children’s Interest Dynamics While Communicating with Robots

The purpose of this study is to develop a communicative robot that estimates other’s mental state, and to elucidate a computational principle of brain dynamics that realizes a mental level interaction. As the first step for the perspective, we had constructed robots that can observe and estimate child’s internal states, such as boring or enjoying. As the experiment, we observed a child playing a card game with a robot that we had developed, and talking with the robot. After the experiments, the children’s mental states were evaluated by other participants by watching videos. The results showed high correlation with an index calculated by a face recognition system in the robot. This result suggests that it is possible for the robot to recognize human interest and to make an action inducing human-like interaction.