Emotional eye movement generation based on Geneva Emotion Wheel for virtual agents

Eye movement plays an important role in face to face communication in that it conveys nonverbal information and emotional intent beyond speech. Our research proposes a computational framework that enables virtual agents to convey different emotional expressions to users through eye movement. We build this framework based on eye movement parameters derived from the Kohn-Canade AU(action unit)-Coded Facial Expression Database as well as real-time eye movement data. We also describe a rule-based approach to generate emotional eye movement based on the Geneva Emotion Wheel. We then present one experiment in which subjects evaluated the emotional eye movement generated by this framework. When our proposed model was employed, the results showed a higher rate of recognition of the agent intended emotion, proving the validity of our approach.