Affective Facial Expressions Using Auto-Associative Neural Network in Kansei Robot ‘‘Ifbot’’

Abstract

The authors propose three methods of enabling a Kansei robot, Ifbot, to convey affective expressions using an emotion space composed of an auto-associative neural network. First, the authors attempt to extract the characteristics of Ifbot‘s facial expressions by mapping them to its emotion space using an auto-associative neural network, and create its emotion regions. They then propose a method for generating affective facial expressions using these emotion regions. The authors also propose an emotion-transition method using a path that minimizes the amount of change in an emotion space. Finally, they propose a method for creating personality using the face.