Effects of a Teacher-type Robot Using a Perplexion Estimation Method on College Students

Kohei Okawa, Felix Jimenez, Shuichi Akizuki, Tomohiro Yoshikawa

Abstract: Educational support robots have received a lot of attention. This research focuses on teacher-type robots that teach learners. Conventional teacher-type robots don’t provide learning support autonomously. Those robots can only provide learning support by the learner’s button operation. In real educational situations, the teacher can guess what the learner is perplexing and provides support by calling out to the learner. The learning environment with the conventional teacher-type robot is distinct from a human-to-human environment. The teacher-type robot’s impression is also reduced. Thus, we considered it crucial to develop a method to estimate the perplexion from the learner’s facial expression. This research defines perplexion as the state of a learner requesting assistance. Based on deep learning, this study built a perplexion estimation method. This method can estimate the perplexion state from the learner’s facial expressions and develop a teacher-type robot. This paper investigates the impression effect on university students of collaborative learning with a teacher-type robot that estimates the learner’s perplexity and provides autonomously learning support. The experimental results findings proposed that using the proposed method for teacher-type robots can achieve a learning environment like that of humans.