Abstract
To realize an electric shoulder prosthesis arm system that can be used without long-term training, we developed an intuitive shoulder disarticulation prosthesis system. The developed system consisted of four degrees of freedom joints and control system adapting a user through machine learning technique and surface myoelectric potential of the trunk. We measured the surface myoelectric potential of the trunk of healthy subjects at multiple points and analyzed by using a principal component analysis to identify the proper myoelectric potential measurement part of the trunk. As a result, it was revealed that the proper EMG measurement site of the trunk is distributed in the chest and back. In addition, as a result of the experiment of verifying the grasping / moving motion of the object by the shoulder artificial arm using the surface myoelectric potential of the chest and the back for controlling the motion of the arm, all the subjects successfully grasped and moved the object within a certain period of time.
