Avatar Interfaces

As a consequence of my thesis research, which became more and more focused on the interrelationships between agents and the people who interact with them, I turned to the question of how people interact with, understand, and relate to their avatars, or autonomous agents which represent them on-line or in virtual environments.  In my project on avatar interfaces, I analyzed the metaphors underlying current avatar technology and proposed new technology based on  different metaphors - read all about it in my Fulbright Project Proposal!

One of the results of this project was my work on semi-autonomous avatars in the Traces system, a collaboration with Simon Penny, Jeffrey Smith, Andre Bernhard, and Jamie Schulte. In Traces, a person interacts in a CAVE immersive virtual environment with his or her own avatar representation, and potentially those of others in CAVEs  connected over a network.  The avatar starts out as a volume model of the user's body movements, which lingers in a 3-dimensional analog to stop-motion photography.  It then gradually becomes more and more independent, spawning particles  which over time develop their own behavior and interact with the user.  Traces provides a rich framework in which to explore concepts of ``semi-autonomous'' avatars and their relationships to human users.

You can check out these mpeg movies of Traces at work:

• The output of the vision system: a 3-D voxel body model, captured in real-time.
• The passive trace (here rendered slower than in real life) streams behind the user and fades away. The user in the CAVE experiences the blue trace behind him; the grean user body model is for your benefit.
• The active trace is formed like the passive trace, but develops using a 3-D cellular automaton model.
• The behaving trace flies off the user's body movements, then flocks around the CAVE and interacts with the user's movements.

Back to my work page.