What happens when a child learning Math online has not just a monotonous tutor, but has 2 cute robots, one sitting beside them and another inside their screen? This project, part of my ongoing PhD research at Osaka University, set out to explore exactly that.
We designed a multi-modal robot system using CommU, the tabletop humanoid robot developed at Osaka University in collaboration with Vstone Co., Ltd. Japanese children aged 9–10 used Khan Academy to practise Grade 3 Math, supported by two versions of CommU working together: A physical robot on the desk providing praise, motivation, and guided break activities, and a virtual CommU appearing on-screen to help children stay focused on the learning material. The system adapted in real time, tracking correct and incorrect answers, triggering dynamically timed breaks, and mixing verbal and non-verbal encouragement to keep children engaged without overwhelming them.
We compared this embodied experience against a voice-only condition, where the same dialogues, break schedules, and encouragement were delivered through hidden speakers. Both groups completed the same Math exercises, and we measured their Math performance, focus, and emotional expressions through tests, questionnaires, and detailed video analysis.
The results told a clear story: While both conditions helped children stay on task and complete Math problems more efficiently, the children interacting with the multi-modal robots were visibly happier. They smiled more, laughed more, and showed genuine excitement — reactions that were absent in the voice-only group, where boredom and disengagement were far more common. Embodiment showed a promising trend toward improved Math performance, though not statistically significant in this short pilot, but it meaningfully changed the emotional quality of the learning experience. This distinction matters: Sustained positive emotions are key to building children's long-term confidence in subjects like math.