Max - A multimodal assistant in virtual reality construction

Abstract

In the Collaborative Research Center SFB 360 ``Situated Artificial Communicators'' the anthropomorphic agent Max is under development. Max inhabits a CAVE-like Virtual Reality environment where he is visualized in human size. Engaged in cooperative construction tasks, the human user meets a multimodal communication partner that employs synthetic speech, gaze, facial expression, and gesture. Combining deliberative and reactive conversational behaviors with the capability to initiate assembly actions, the agent guides the user through interactive construction procedures. This paper describes the overall architecture of Max with a focus on dialog and interaction management and the generation of synchronized multimodal utterances.