Self-learning system of facial expression for intelligent robot through interaction with human

We developed an intelligent robot which understands human facial expressions and behaves according to their facial expressions, through the interaction. We implemented a self-learning system onto the robot for understanding the expressions. The self-learning system models the process that a baby learns his/her parents' facial expressions. Equipping the robot with a camera, the system can get face images and equipping the CdS sensors on its head it can get the information of human action. By analyzing the information of these sensors, the robot can obtain a feature of facial expressions. When the equipped sensor detected human actions, the system extracts facial feature from camera images, and memorizes the feature of facial expressions by carrying out repetition learning. Consequently, by comparing reserved feature with input facial image, the system can recognize facial expressions without using the information of human actions. We implemented the system onto our robot and investigated its usefulness. Through the self-learning system, the robot became to recognize human facial expressions, and could behave according to person's expression.