User Guidance in Virtual Worlds

Abstract

We introduce a software architecture which supports user guidance in virutal worlds. The user guidance is realized with task-oriented agents communicating with users through a command-based dialogue interface. The agents use an application independent toolbox, the hyper-renderer, for evaluating information that is calculated during the rendering process, but usually not provided for usage outside the rendering program. As an application, we describe a prototypical exploration agent which guides visitors of a virutal building to any destination inside the building. The visitor is navigating independently, while the agent presents sign-posts at apporpriate locations. In our implementation, we used symbolic visibility preprocessing. Example frames of an interactive walkthrough are included.