Generating Interactive Facial Expression of Kansei Robots Using Simple Recurrent Network

Abstract

To improve face-to-face interaction, we propose a model for generating interactive facial expressions using a simple recurrent network (SRN). Conventional methods using built-in facial expressions only make static face changes, so users sense incompatibility and monotony from those expressions. Therefore, generating expressions by synthesizing designed facial expressions dynamically is required. In this paper, we developed a robot having dynamic face changes using our model and compared face changes of our model with those of the conventional model. The results suggested that our model can increase the diversity of interaction.