Human-Robot Interaction based on Facial Expression Imitation

Abstract

Mimicry during face-to-face interpersonal interactions is a meaningful nonverbal communication signal that affects the quality of the communications and increases empathy towards the interaction partner. In this paper we propose a facial expression imitation system that utilizes a convolutional neural network (CNN). The model was trained by means of the CK+ database., which is a popular benchmark in facial expression recognition. Then, we implemented the proposed system on a robotic platform and investigated the method's performance via 20 recruited participants. We observed a high mean score of the participants, viewpoints on the imitation capability of the robot of 4.1 out of 5.