Study On Interactive Intelligent Virtual Agent And Its Emotion

Intelligent virtual agent is an important part of 3D graphical simulations. It can be used in many fields, such as film, game, E-Learning, military simulation, et al. To simulate the real world and enhance the reality and immersion of the virtual environment, how to increase the intelligence and believability of intelligent virtual agent has been the key problem that most researchers focus on. We study on interactive intelligent virtual agent and its emotion, especially the methods of building cognitive process, model of generating emotion, and methods of integrating user direction into the intelligent virtual agent's behaviour. Our study includes several aspects as followed:(1) Based on ethnologists and psychologists' study, this dissertation analyses the important role of motivation and emotion in cognitive process, and then designs an active cognitive architecture. At the same time, the implement and optimization of the virtual vision, motivation, action plan in this architecture are presented.(2) Taken into account the influence of physical needs and cognition, a comprehensive computing emotional model is proposed. An architecture of emotion process is presented to describe the dynamic characteristic of emotional state. The emotion structure with cause events is added to generate an emotion state with certain strength, which then choose the most appropriate emotion behaviour. In addition, a social learning component is presented to improve virtual agents with the ability of adaptability in the dynamic environment. In order to create various personality virtual agents, we also introduce a simple nonlinear function to simulate the influence of personality on emotion. By using multi-channel expression, such as facial expression, gesture and body movement, the internal state is finally transformed onto the concrete behaviour to express the personality virtual agents.(3) Based on the above cognitive architecture, this dissertation makes further study on interaction between users and intelligent virtual agents and then presents an effecitve method that combines the active cognition and the user's directable behaviour. When the user gives an order, the intelligent virtual agent can generate believable behaviours by means of comparing the user's order with its emotion. This method can not only increase the user's interaction and immersion but also create personality-rich intelligent virtual agents.(4) Sythesizing the previous methods, an original animation emotion system is developed. We use experimental tests and human-subject ratings to evaluate our proposed methods and the results show that our system is valid and has a potential future. Based on the products of ethology, psychology, cognition and artificial intelligence, etc, we study on interactive intelligent virtual agents with emotion intelligence, which provided a practical direction to enhance the reality and immersion of the virtual environment. Applied with the theory and technology presented in this dissertation, the intelligence and believability of the intelligent virtual agent are improved. Moreover, the system's preferable practicability and its promising situation have confirmed that the dissertation has a reasonable and farseeing objective.