Happy Companion : A System of Multimodal Human-Computer Affective Interaction
Jia,Lianhong Cai,Sirui Wang,Xiaolan Fu
2010-01-01
Abstract:Emotion is one of the most important aspects in the intelligent human computer inter-actions. The concept of “Affective Computing” refers to the “computing that relates to, arises from, or deliberately influences emotion or other affective phenomena”. The research on realization of emotion comprehension and expression has been a hot topic. In this paper, we proposed a novel human-computer affective interaction system, called “Happy Companion”. We adopt the PAD model with three dimensional: pleasure (i.e. valence, evaluation), arousal (i.e. activation, activity), and dominance (i.e. power, potency), as the description of human’s emotional states. By integrating relevant technologies, such as PAD quantification and annotation, emotion detection and analysis from speech and video, 2D/3D facial expression synthesis, et al., the happy companion presents a virtual guider who can detect your emotional states and respond expressively through audiovisual interface in real time. Under her guide, you can participate four basic tasks to regulate your emotions, including Breathing Relaxation, Muscle Relaxation, Facial Expression Learning, Joke and Game. The design and study of “happy companion” have made a try to create a system of multimodal human-computer affective interaction.