Development of EMG-FMG based Robotic Prosthesis with New Vibrotactile Feedback Scheme

Abstract

In this paper, the development of EMG-FMG-based forearm prosthetic arm with new vibrotactile feedback method is proposed. Existing electromyography robotic prosthesis cause dissatisfaction due to malfunction due to sensor recognition error. The EMG signal used to drive the prosthesis has information directly related to muscle activation but has the disadvantage of outputting different measurement results when the electrode position or sensor is moved, muscle fatigue, and severe noise in the skin or sweat. As a replacement method, the FMG signal is used, which is relatively stable and resistant to moisture or sweat, but has a limitation that it relies on volume changes during muscle contraction and relaxation, rather than measuring muscle activation directly. The combination of EMG and FMG can not only compensate for the problems that occur in single sensing, but also reduce malfunctions due to sensing errors, and is expected to improve overall performance in prosthetic control.

In addition to this increase in control performance, one of the important factors that need to be developed is the application of prosthetic feedback. In fact, amputees find it more functional and easier to receive sensory feedback through additional devices than simply controlling the prosthesis. The Vibration Tactile Feedback method provides vibration feedback of tactile signals felt through the prosthesis. Ideal for prosthetic applications in terms of ease of use, low power consumption, light, and small size. However, vibration motors used for conventional vibrations have limitations that are slow to respond and difficult to control frequency and vibration intensity. Also, the rigid body and thickness make it difficult to wear, causing inconvenience to amputees.

To solve these problems, in this paper, we propose a new method of providing feedback by piezoelectric vibrotactile actuators by sensing tactile sensation by using capacitive knit data glove in a robot prosthesis using EMG-FMG knit data glove as an input device.

First, the structure and measurement method of three sensors used to control the feedback of the prosthesis were proposed. The three sensors consist of an EMG-FMG knit band sensor developed to measure EMG-FMG signals used for prosthesis, a capacitive knit data glove developed to detect the tactile sense of prosthesis, and a piezoelectric vibrotactile actuator developed to give feedback on detected tactile information through vibration.

Second, we analyzed the motion recognition method and performance evaluation based on EMG-FMG. The motion was classified using the threshold-based method and the machine learning-based method, and the performance was analyzed to select an appropriate classifier for prosthetic hand control. In the prosthetic feedback control, EMG-FMG signal classification was performed using the double-threshold method and SVM.

Thirdly, a biofeedback control method of Robot Prosthesis applying a new vibrotactile feedback method was proposed. Here, the 850g HYU forearm and EMG-FMG Knit Band Sensor are used to detect and process muscle signals, and to identify the intention to create and manipulate the corresponding prosthesis in the feed-forward control loop, capacitive Knit data glove The sensed tactile information is mapped to it using a piezoelectric vibrotactile sensor and includes a sensory feedback loop for feedback to a person.

As a result, the performance of the proposed method was verified through two experiments, and as a result of the experiment, when there is biofeedback in the use of the prosthetic arm, it was confirmed that the task completion time was shortened and errors during the operation were reduced. Finally, future studies will aim to evaluate practicality and safety by conducting experiments on amputee wearing a developed prosthesis.