The Research Of Emotion Model Based On Particle System

Artificial Emotion has broad prospects and potential economic value in a harmonious human-computer interaction, education, anthropomorphic robotics, and entertainment. Many universities and companies have set up some research groups of artificial emotion, such as Massachusetts Institute of Technology, IBM, SONY Corporation and so on. The research of emotion modeling has become a hot topic of scientific computing and psychology in recent years.Based on C.Breazeal's 3-Dimension emotion space and OCC model, combined with particle system and active field, this dissertation builds an artificial emotion model. In emotion space the particle system denotes the status of emotion model. Those outside exciters create the active fields in space respectively. The potential energy which was given out by several active fields impacts the movement of particles system. The model outputs a multi-dimension vector continually, which describes the level of emotional activation at a time. If one or more the emotional dimension is more than the activation threshold, it means that the emotional model has these types of feeling at the moment. For the above-mentioned model, this dissertation implements a system of Artificial Emotion. Based on the Facial Action Coding System and Candide grid model, the output of system will be in the form of facial expressions accordingly.Compared with the research of Artificial Emotion domestically and abroad in recent years, the main contribution of this dissertation is as follows: in accordance with characteristics of human emotion such as attenuation, nonlinearity and constancy, particle system is introduced creatively to reflect the changes of emotion model in emotion space. Consequently, for background emotion, model may be in different emotional states with same external stimulation. The model can not only denote the impact of several external stimulations in emotion, but also reflect the potential effect of the events which occurred before. Meanwhile, multi-dimensional output makes that algorithm possible, by which dynamic facial expression can be generated.