Affective and Attentive Interaction with Virtual Humans in Gaze-based Settings

This thesis investigates the affective and attentive gaze-based interaction with virtual humans and will answer related questions to affect and attention and the challenges facing the generation of an anthropomorphic interface for human-machine interaction. One focus is on recommendations for the generation of gaze aware interfaces, the control and perception of virtual agent's facial expressions and an interactive gaze model for the face-to-face interaction with a virtual agent. Our focus is not on the analysis of the semantics, but on the socio-emotional aspects of such a conversation. This thesis gives a general view on the different functions of gaze and eye contact in human-human interaction and related work of human-agent interaction and how interfaces with explicit and implicit gaze interaction could look like and reveal the challenges of such interfaces. While facial expression control can be quite complicated, this thesis demonstrates how facial expression control can be simplified. The investigation of the perception of facial displays inbetween humans and virtual humans is part of this thesis. It investigates how humans perceive variations of the facial display of a virtual human. One part deals with how humans perceive the emotional component of the combination of affective facial expressions, head direction and eye orientation. Another part investigates whether the head direction and eye orientation influence the perception of the personality of the virtual human. Finally, mutual gaze plays an important role to establish attention with an interlocutor. This thesis deals with a gaze model that is aware of a user's current gaze with the help of an eye tracker. It analyzes where a user is looking and if the user is looking into the eyes of a virtual human to establish mutual gaze. It investigates if a human recognizes the interactive gaze model in a non-verbal setting and whether a human recognizes it in a verbal setting