Computational Emotion Model for Virtual Characters

Abstract

Computer game is becoming a new computer application field, its one of key technology is to set up intelligent virtual characters. Emotion is of the highest importance in modern computer games. The essence of computer game lies in waking a user’s emotion experience; modeling virtual character’s emotion is the key goal in computer game. David Freeman pointed out in his best-seller book “Creating Emotion in Games”(Freeman, 2003) : the revolution in the future of computer game is not the technology, but to create emotion experience. Truthfully, many directors of computer game products have already realized that the contents must be full of emotion experience. Computer game industry needs more clever virtual characters with an intelligent model. Intelligent virtual character is a new research field that integrates artificial life and computer animation together. Artificial life is the research field that tries to describe and simulate life by set-ting up virtual artificial systems with the properties of life. We can get more understanding from the developing history of computer animation. Early computer animation only includes shape and movement of a geometry model, it is very difficult to draw complex natural landscape, and artificial life can help to solve these problems. In a general, an artificial life model is based on bottom-up strategy. Emergence is the key concept of artificial life. It means a complex system is from the simple location interactions of individuals. Another key concept of artificial life is adaptation, which means evolution. In 80 years of the 20th century, many models of computer animation were presented, such as particle models, L-system, kinematics and dynamics, facial animation, etc. In 90 years of the 20th century, artificial life influenced the development of the computer animation greatly (Tu, 1994); Funge presented the cognitive model for computer animation (Funge, 1999), On the basis of the Funge’s a cognitive model for computer animation, we can illustrate a virtual character’s hierarchy of computer animation in Fig 1. In order to create believable characters, people hope to set up computational emotion models for virtual characters. There are a lot of relative researches on virtual character and emotion model; we only introduce part of them. Badler et al. use finite state machine to control a virtual character's behavior, personality characteristic was expressed by locomotion parameters (Badler, 1997), they also built a system called Emote to add personality and emotion for virtual characters (Chi, 2000), Ball et al. proposed a Bayesian network-based model of personality for speaking