A Robot for Reconstructing Presentation Behavior in Lecture

In universities, lecturers often use presentation slides to present their lecture contents with non-verbal behavior involving paralanguage, gaze and gesture, which is so important for promoting learners' understanding. However, it is not simple for the lecturers to appropriately control non-verbal behavior depending on the lecture contents. This paper proposes a lecture robot, which substitutes for lecturers, and which reconstructs their non-verbal behavior conducted in their lecture. Towards such reconstruction, we have designed a model of presentation behavior in lecture. This paper also demonstrates a lecture robot system, which detects insufficient/inappropriate behavior to reconstruct and reproduce by following the presentation behavior model. In addition, this paper reports a case study using the system with 31 participants, whose purpose was to compare the video-recorded lecture conducted by a lecturer and the lecture reconstructed by the robot system. The results indicate that gaze, face direction, and pointing gesture for keeping and controlling learners' attention reconstructed by the lecture robot are more acceptable and understandable. It is also suggested that the robot could promote their concentration by means of eye contact with them, and that it has a potential to promote their understanding of the lecture contents.