Development of a 3D-Printed Myoelectric Arm Prosthesis with a Rotational Wrist Joint

Abstract

Prostheses can be highly expensive and often require a long manufacturing time, leading to limited accessibility to prosthetic devices for impaired individuals. This thesis aims to develop a low-cost, 3D-printed arm prosthesis controlled by electromyographic (EMG) signals. A MyoBand is attached to the residual limb to record eight EMG signals, which are used to differentiate between various hand gestures. The prosthesis features intuitive rotational control of the wrist which has not yet been implemented in other, similar devices. Therefore, the following research questions are asked: (1) To what extent can intuitive rotational movements of the hand be recognized using a MyoBand and differentiated from other gestures? (2) Can the data collected by the MyoBand be used to enable proportional and simultaneous control of the rotational movement along with other hand gestures in a 3D-printed arm prosthesis? In order to answer the research questions, this thesis was conducted through the following steps: First, an appropriate 3D model was selected, which was adapted in order to be applicable as a prosthesis. Second, the required hardware components, such as motors and a microcontroller, were selected to match the design specifications and performance needs of the prosthesis. Third, the individual parts of the prosthesis were 3D-printed and assembled, and the hardware components were integrated. Fourth, the software for the control system was developed using MATLAB and Simulink. The control system comprised two programs: one for capturing various gestures, including flexion, extension, pronation, supination, and relaxation as a baseline, and another one featuring a Linear Discriminant Analysis (LDA) classifier for recognizing these gestures and controlling the prosthesis accordingly. Fifth, the control system was tested using a virtual prosthesis to validate its functionality, followed by a pilot test with the physical hardware. The pilot test was conducted to evaluate the LDA classifier’s gesture recognition and to determine whether the 3D-printed prosthesis could be controlled based on the recognized gestures. The results demonstrated that the system was capable of distinguishing between different intuitive gestures, including rotational movements. Furthermore, the motors of the prosthesis could be controlled proportionally to the user’s muscle strength, allowing for a precise control of the prosthesis. Interestingly, the pilot test also revealed that the position of the arm influences the gesture recognition capabilities of the LDA classifier, with the suspended arm leading to higher accuracy compared the flexed arm.