Muscle synergy of biceps brachii and online classification of upper limb posture

摘要

For someone who has suffered from a partial arm amputation, muscular surface electromyographic (EMG) signals are usually used to control a myoelectric prosthesis. To learn how to satisfactorily control the prosthesis, software programs can be very useful. We present here a program, of which 5 EMG signals collected across the biceps brachii are decoded to produce signals that either make a simulator replicate the arm posture or control the position of a small humanoid manipulator. In the program, following a phase where muscle synergies are extracted from a training trial, the learned features are then used to classify the following arm postures taken by the subject. The mean classification performance for 56 different two-class paired arm postures is 94.9% for 2 normal subjects. Following further testing with normal and amputee subjects, the system could eventually be used in rehabilitation centers where upper limb amputees want to use a myoelectric prosthesis.