CADIA Populus is an interactive 3D environment for social behavior simulation. It is geared toward the development of autonomous behavior for agents and avatars in multiplayer game environments.
CADIA Populus combines full online multi-player capability with clear visual annotation of the social environment in terms of the theoretical models being used to inform automated avatar and agent behavior.
A flexible and powerful framework for automation is provided with the tool, part of which is inspired by the OpenSteer
model which allows steering behaviors to be activated and deactivated.
The agents and avatars are human ’clay’ figures with articulated necks, drawn to scale in the environment for accurately reflecting distances. Users can drive them around and start conversations between them using the built-in text chat.
A single developer can generate any number of figures and switch between them at will, which is perfect for manipulating the social situation.
Each automated entity has access to the social environment, through what we call a social perception interface, can be used to trigger behaviors. Examples of social perception include how many individuals are within a certain proximity range (intimate, personal, social and public) or whether the avatar has stepped within the domain of an active conversation according to Kendons F-Formation model.
The perception of social and public space has a blind cone behind the avatar, of respectively 90 and 150 degrees. Both perceptual visualization and the set of active behaviors can be shown inside the environment.
The tool is written in Python and uses the Panda 3D game development library from CMU and Disney, and incorporates the PhysX physics engine from NVidia