Physical therapy massage robot system with human motion tracking and finger-pressing force control
Naoya Harada, Michiteru Kitazaki, Ryosuke Tasaki
Abstract
Massage is a technique that heavily relies on the human senses. The application of massage to the human body is subject to many uncertainties. These include variations in the recipient's body shape and posture, as well as fluctuations in their condition, like movements and changes in stiffness. Three elements are needed to solve this problem: understanding the massage technique, determining the position to touch, and adapting to the recipient's movements during contact. The parameters of massage technique are elucidated by analyzing the massage movements of a therapist. The skeletal points of the recipient are estimated by the RGB-D camera to determine the 3D massage position of the recipient. The force feedback control, focusing on periodic input, adapts to the recipient's periodic body movements. Experiments demonstrated that the system adjusts finger-pressure position and force according to subtle recipient movements and muscle stiffness while lying down. The physical therapy massage robot autonomously massages while adapting flexibly to the recipient's condition, similar to a licensed therapist. The physical therapy massage robot system will contribute to human-robot interaction, such as providing interactive or social touch between humans and robots.