W. Lewis Johnson
Director, Center for Advanced Research in Technology for Education (CARTE)

Expression, Intention, and Context in Animated Agents

This presentation examines the problem of how to control behavior in 
virtual actors, i.e., autonomous agents that can play roles in 
interactive dramas, games, or learnin environments.  Unlike most 
animated characters whose behavior is scripted in detail ahead of 
time, virtual actors need to be able to be able to choose what to do 
and say in real time, based upon the situation, the actions of other 
virtual actors, or the actions of human audience  members or 
participants. Furthermore, the intent underlying the actions of the 
virtual actors need to be clearly understandable by the human 
observer; this requires that the virtual actor's behavior have 
expressive qualities, just as human acting does, and needs to be 
performed so that the audience can see it and it doesn't interfere 
with the audience's understanding of the actions of other characters.

I will argue that virtual actors require more than an abilty to 
replay expressive motion, acquired through motion capture or other 
means. Rather, they require an explicit model of the intent of the 
actor (its cognitive and emotional state), a model of the context in 
which the behavior takes place, and an ability to generate behavior 
in a manner that takes the intentional and contextual factors into 
account.
Important contextual factors include the social context (the social 
relationship that the character has with other characters or 
people), the staging context (how the action is organized on the 
computer display, considered as a virtual stage), and the dramatic 
context (the the function that the action plays in the drama, and 
its relationship to other actions that occur before or after).  This 
approach is closely analogous to human acting technique, 
particularly in stylized dramatic genres such as opera, where actors 
deliberately link dramatic intention and dramatic gesture.  These 
points will be illustrated by virtual actors that we have developed 
and are developing at USC's Center for Advanced Research in 
Technology for Education and Institute for Creative Technologies.