An Interactive Robot Lecture System for Attention and Understanding Recovery

Abstract

In lecturing, lecturers need to control the attention of learners to maintain while monitoring their learning states. We have claimed that a robot as lecturer can properly conduct nonverbal lecture behavior to control learners’ attention and enhance their engagement compared with human lecturers. Although attention control is possible in shorter lectures, it would be more difficult for the robot to maintain learners’ attention in a longer lecture. This prevents them from becoming aware of an important part of the lecture contents to understand. In this work, we have been developing an interactive robot lecture system that interacts with learners to recover their attention and understanding when they overlook the lecture contents. To recover learners’ attention, the system first estimates their states of attention and understanding from their posture and gaze during lectures. Second, the system reconstructs a lecture scenario to generate an interactive lecture behavior, which combines non-verbal behavior about pause, walk, and repeat with paralanguage. The interactive behavior is implemented using a humanoid robot NAO. Third, the system provides questions about the slide contents for recovering their understanding when they lose their attention in spite of attention recovery.