AI-Enabled Framework for Augmenting Upper Limb Rehabilitation With a Social Robot

This paper proposes an innovative framework that is designed to strengthen human upper limb rehabilitation through the social robot TIAGo. Such a system is designed as a complementary aid used to encourage the patient during a training session by providing contextual feedback based on the comparison of his/her performance with respect to the expected motion. The ground truth is acquired in an initial stage via an accurate motion capture system, where the physiotherapist teaches the targeted motions to the robot. The result is an archive of rehabilitation exercises that the robot can deliver to the patient, the intent of which is to replicate them under the robot supervision. The approach followed in this work is mainly focused on re-targeting the human trajectories into the robot: first, by extracting the marks of the human arm joints with a skeleton tracking software based on the robot’s camera images, then, by operating a scaling and an optimization of the robot arm motion employing the robot’s kinematic model, with the aim of improving the naturalness of its movement.