Learning in an intelligent agent augmented multi-user virtual environment

Abstract

Agents, or computer generated characters, are an integral part of virtual worlds. They are particularly important in virtual learning, where their interactions with students foster more effective and situated learning experiences. Currently, agents in virtual learning worlds have a number of limitations. This project seeks to study these limitations and implement theories to overcome them. Agents in virtual environments are modeled on a goal-based model, wherein they have a limited set of goals and all their actions are directed towards achieving those goals. While this may ensure that a goal is achieved, it also limits the agent’s actions in response to users. This project looks at implementing the FALCON model to make agent-student interactions more dynamic and realistic. Agents in virtual worlds currently also lack in emotive capability. Especially in the educative context, there has been very little attention paid to the importance of agents understanding and displaying emotions. This project will further look at the OCC model for emotion-cognition, and its implementation with the virtual world. Lastly, this project examines a teaching concept called Productive Failure, which proposes that unstructured environments may help in student learning in the long term. The project investigates how to employ this concept in the virtual world, and its possible impacts.