Intelligent agents in virtual world

Abstract

An intelligent agent is able to receive input and act in a certain manner. It is also capable of learning to a certain degree and can represent knowledge. The study of knowledge learning and representation in intelligent agents can benefit systems deploying autonomous/intelligent agents in information gathering, representation and decision making. This purpose of this project is to carry out investigation and implement an intelligent agent in a virtual world. The scope of this project is confined to the intelligent agent being able to query and collect information about various entities, and store them meaningfully in its knowledge base. A form of pattern/template matching learning called Fusion Adaptive Resonance Theory is also deployed in the learning process, and it will support the intelligent agent in providing a system for concept formation knowledge retrieval via template matching. The techniques, approach and design used to achieve the objectives of this project will be discussed in detail in the following chapters. In conclusion, it is indeed feasible to deploy an intelligent agent which can roam a virtual world, gather information and present the information to human users. However there are timing issues and limitations in the development of the virtual world which have to be overcome. There is also future potential for the agent implemented in this project to be integrated or complemented with other agents to provide a better gaming experience for the user.