Published: 30 September 2007

Development of threshold based EMG prosthetic hand

A. S. Arora1
Deepak Joshi2
Views 0
Reads 0
Downloads 1027

Abstract

There is a real need of EMG (Electromyogram) based prosthetic hand for the amputee which should be economical as well as reliable. The cheap prosthetic hand available in market works passively. In those cases the patient does not feel the feeling of natural human hand. EMG based prosthetic hand provides the amputee feeling of natural human hand. The work that has been discussed here is to develop a prosthetic hand with one degree of freedom. The two motions developed were open and close. Most of the work is done at electronic level. The main work was to acquire the noiseless EMG signal and further to convert it into control signal for prosthetic hand, after suitable processing. For classification a threshold based technique has used rather than any classification technique like Artificial Neural Network (ANN), Fuzzy Logic and Genetic Algorithm (GA). It was tried to use the minimum hardware, without making any compromise with performance. It was done so, to achieve the target of developing a economical and reliable prosthetic hand. The threshold value used was variable and was controllable from outside by just varying the knob of potentiometer. This adds an additional dimension for tuning the device and scope to adjust the threshold according to muscle activity of subject. So the same prosthetic hand can be used by different amputees by just changing the threshold values only. The mechanical hand was having only two fingers to grasp the objects. The work was also extended to develop the frequency based Prosthetic hand. The scheme was to find out the frequency bands where the amplitude of open and close motions is different. The FFTs (Fast Fourier Transform) of EMG signal were calculated in MATLAB. The DSO (Digital Storage oscilloscope) was also having the facility of displaying the FFT of signal. It was found that there is certain possible frequency band which classifies the open and close motion of hand

About this article

Received
21 March 2007
Accepted
18 June 2007
Published
30 September 2007
Keywords
Electromyogram
Artificial Neural Network
Fast Fourier Transform