Human-Robot Interaction Based on Facial Expression Recognition Using Deep Learning

In recent years, many robots for the purpose of communicating with people have been developed. Such a robot is required to have human interaction and communication ability. In order to perform the interaction naturally, the nonverbal communication such as human facial expression and body movement is important. In this research, we propose a method to classify emotions from human face images by deep learning and generate a robot emotional reaction by Markovian emotional model. Here, we perform to learn human facial images with various emotions using CNN (Convolutional Neural Network) which is a kind of deep learning, and recognize human emotions from facial images in the human interaction. Based on the human emotion obtained by deep learning, the robot returns its emotional behavior to the human. In this research, we executed the interaction experiment using an real communication robot and this result is also reported in this paper.