Motion-Copying System Based on Modeling of Finger Force Characteristics Using Upper Limb-EMG

In recent years, motion-copying systems that store and reproduce human motions have attracted much attention. In the conventional method, the motion is stored using a motor, which affects the original task. In this study, we focused on the relationship between electromyography and force in order to realize unconstrained, non-contact force measurement. In this paper, we modeled the relationship between the force of pressing a force sensor with a fingertip and the myoelectric potential of performing an action by using the elemental description method, which is one of the system identification methods with easy physical interpretation. As a result, an accuracy of 0.260 N, the least squares error, was obtained. In addition, we conducted on copying and reproducing the motion of finger using this model. Although the accuracy of force estimation was low, we were able to estimate the force with the same accuracy. In the future, we aim to improve the accuracy of the estimation and to measure the force using only the myoelectric sensor without the force sensor.