Adaptive Human Behavior Avatars (AHBAs)


Click the play button on the image above to view the demo video.

IDI is currently developing automated BOTs for use as non-player characters (NPCs) in virtual training environments. The BOTs are implemented using a technology called Dynamic Decision Networks (DDNs), which use Bayesian networks to integrate inputs to the NPC from the player and influence diagrams to guide NPC decisions in order to adapt the NPC decisions as the training scenario unfolds.  In the current implementation, players send text statements and questions to the DDN-based NPC. The NPC then determines how warmly and truthfully to respond and provides the appropriate textual response. These responses range from helpful statements of truth to outright lies or ending the conversation.

For a copy of the project summary please click here.

 

AHBA Editing Tool