Research On Agent Technique For Intelligent Virtual Environment

Intelligent Virtual Environment (IVE) is the integration of Virtual Environment and Artificial Intelligence. The objects, inhabited in the environment have lives that can be embodied by intelligent agent. This research will contribute to the development of a new generation of virtual environment with the realistic behavior, it's also supports the interactive collaboration work in distributed virtual environment.The dissertation focuses on the agent technique in Virtual Environment, including the interactive behavior of physical agent for virtual environment, the personalized recommendation algorithm based on cognitive agent, the agent autonomous behavior simulation in virtual environment and the design of the autonomous human perceptional model.On the interactive behavior of physical agent for virtual environment, firstly, this thesis proposes the interactive hierarchy of agent in distributed virtual environment, we design the interactive algorithm based on this frame. Secondly, in order to solve the conflict between agents, we adopt the nearest neighbor algorithm to match the instance of agent. Thirdly, the thesis researches the model for generating and animating groups of virtual agents. Finally, we implement the collaborative virtual environment based on multi-agent to improve the proposed model and methods.On the personalized recommendation algorithm based on multi-agent, the thesis proposes the method to integrate the cognitive agent with virtual environment, to solve the problems of the single processing behavior and sensibility expression for the traditional cognitive agent. The personalized recommender part focuses on the personalized recommendation algorithm. We use data mining technique, especially K nearest neighbor, association rules, and content-based recommendation algorithm to improve the accuracy and individuation of products recommendation.On the human autonomous behavior simulation in virtual environment, the thesis presents a model of human autonomous behavior. Firstly, we propose the human autonomous behavior selection, behavior hierarchy of human, selection mechanism of autonomous behavior, and the implementation of autonomous behavior programming. Based on the motivational model, we develop a new mechanism of behavior selection, which can select the appropriate behavior to interact with object by using the human internal motivation and environment information. We also design the human behavior hierarchy via behavior classify. We implement the selection mechanism of autonomous behavior with the inhibition and fatigue model. Based on the external and internal stimulative information, we implement the autonomous behavior programming with producing rules.On the design of the autonomous human apperceive model. In order to apperceive the environment surrounding the human, we adopt the human perceptive model to collect and process the environment information. The thesis also proposes the queue virtual memory module to solve the object information occurred, the Bilinear Interpolation algorithm of motion control module to implement the human animation.