Development of a Learning Companion Robot with Adaptive Engagement Enhancement

Abstract

This research aims to develop a learning partner robot that can adapt its interaction to each learner's individual reactions to enhance learner engagement. Engagement is a mental state that positively influences learning and is an essential element that supports learner independence through immersion in the learning process. However, no established methodology exists to enhance learner engagement with proper support in the individualized and isolated learning process. This research focuses on a learning partner robot as a learning companion in self- directed learning. To realize the partner robot, we implemented the engagement estimation architecture from the learner's facial images during learning, then designed a robot interaction model to enhance learner engagement according to their engagement states, and extended the interaction model to update the robot's strategy according to the learner's response to the interactions. We conducted a comparative experiment with 20 graduate student participants with and without robots. The results indicated that the average engagement during learning was significantly higher in the with-robot condition, and the satisfaction with the robot interaction was highest in the last 1/3 of the learning period.