Abstract
A purpose of this study is to develop motion assistive device for an upper limb. Motion assistive device should be controlled based on the model holding the characteristics of musculoskeletal individual to assist in safely. In this paper, measured the EMG signal and the angular velocity when were the wrist palmar flexion, dorsiflexion and the elbow flexion, extension. It was proposed a method based on the input EMG and output wrist, elbow angle to derive a MIMO-NARX model, estimates the wrist, elbow angle. MIMO-NARX model can represent a multi-input multi-output system nonlinear, and retain the characteristics of each muscle. Using the identified MIMO-NARX model, it was confirmed that it is possible to estimate the wrist, elbow angle with high accuracy, and coefficient of the MIMO-NARX model correspond to the size of the muscle mass. Fit rate of wrist angle was 99.72%, elbow angle was the 99.52%. Value applied to the flexor carpi ulnaris muscle was -15 to 10, extensor carpi radialis longus muscle was -10 to 2, biceps brachii muscle was -50-200. From this, it was considered that it is possible to evaluation of muscle mass to have of individuals from the coefficient of MIMO-NARX model.
