A model to output optimal degrees of emphasis for teaching motion according to initial performance of human-learners-an empirically obtained model for robotic motion coaching system

Slight differences between a learning target motion and motions performed by learners are not easy for learners to recognize, when some adjustments are needed by learners for learning motions. To assist human-learners recognize the slight differences in motions, not only symbolic expressions but also motion display plays very important role. In the previous work, we proposed a motion coaching system that dynamically modify displaying motions according to performance of human-learners. The dynamically modified displaying motions are such that they emphasize a learning target motion in direction to supplement missing elements of performed motion by learners compared to a learning target motion. However, degree of emphasis of the dynamically modified motion are given by an preliminary experiment. The robotic coaching system should be able to make decision for the degree of emphasis dynamically, in order to assist human-learners to recognize slight differences between a learning target motion and performances of human-learners. Thus, in this paper, we conduct experiments where the robotic system coaches humans motions with many different degree of emphasis and record resulting performances of human-learners for each cases. Analysis of experimental results and obtained model suggest that an optimal degree of emphasis for dynamically modified displaying motions can be decided according to initial performance of human-learners.