Comparison of Facial Expression Sensors and Skin Potential Response Sensors for Emotion Estimation by Communication Robots

A communication system that quantifies human emotions and provides feedback control was fabricated for a communication robot. The robot can provide suggestions to users and relay transmissions during human-to-human communication. For instance, if the user is excited, the robot can calm them down, and if the user is sad, the robot can cheer them up. The reliability in estimating emotions from facial expressions of a proposed sensor was compared with that of galvanic skin reflex (GSR) and skin potential response (SPR) sensors, which have been reported in several papers. A GSR sensor can measure tension and agitation in humans. The sensor estimating emotions from facial expressions can measure five facial expressions: “Happiness”, “Surprise”, “Anger”, “Sadness”, and “Neutrality”. Furthermore, the method is non-contact and suitable for practical use because it uses images. As the emotion-estimated values are output, they can be used without any privacy issues. The changes in emotions toward subjects using an entertainment robot that can eject soap bubbles were evaluated. An experimental chamber was created to prevent disturbances and measure only the effects of the soap bubble ejection and cessation events. This study reports that SPR sensors and emotion estimation sensors can be handled similarly.