AFFECTIVE ARM AND HAND GESTURES IN ANIMATED PEDAGOGICAL AGENTS

This research focuses on making animated pedagogical agents more emotionally expressive to facilitate students’ learning process and create a true connection between the agents and the students for effective learning. The author of this thesis focused on applying affective hand and arm gestures that were derived from the Geneva Multimodal Emotion Portrayals (GEMEP) core set of affective expressions on top of animated lecturing agents. The GEMEP database includes 145 video recordings of French actors portraying different emotions. Only arm and hand gestures were extracted from the database and were applied on top of lecturing animations of animated pedagogical agents. 131 participants participated in an online perception study where they were asked to watch different animations of six basic emotions (anger, joy, sadness, disgust, fear, and surprise) displayed by male and female characters. Results of the perception study showed that only one angry expression by the female character passed the 75% bar of recognition rate, while the remaining expressions showed a low recognition rate (1.5% -64%). The author also examined whether levels of emotion arousal and valence were correctly identified by the students. Participants’ identification of correct levels of arousal and valence was more accurate than recognition of Ekman’s basic emotions but was still not sufficient for further statistical analysis on validating hypotheses of research on observing how gender of participants and characters influence the emotion perception. Overall, significant reasons were identified that influenced the low perception rate among participants, which are discussed in the conclusion.