Neural network based classification of arm emg signal for controlling of artificial hand

Artificial Neural Network is the technique to mimic of human neural network, the machine having algorithm of neural network can learn as human learning. The recent researches focus on the controlling of artificial hand movement by the source of arm electromyography (EMG) signals. The acquisition of

Project Title

Neural network based classification of arm emg signal for controlling of artificial hand

Project Area of Specialization

Robotics

Project Summary

Artificial Neural Network is the technique to mimic of human neural network, the machine having algorithm of neural network can learn as human learning. The recent researches focus on the controlling of artificial hand movement by the source of arm electromyography (EMG) signals. The acquisition of EMG signal is done by analog emg sensor.

Five subjects will perform five hand movement (hand open, hand close, wrist flexion, wrist extension, victory) in two different elbow positions (0, 90degrees) the elbow angle will be measured by goniometer.

The signal will pre-process and noise will remove from it. Then feature will extract out from it, feature like zero crossing, mean, variance and slope sign change. After achieving performance by using these features more features will be used for better performance. We have the capacity of using many different features for the best performance of the Artificial Hand.  After feature extraction we use different classifier to train the neural network. 

 Neural network will train then we feed it into the artificial hand that will help the trans-amputee patients to move their hand irrespective of their elbow position. The scope of this work in not only limited to the trans-amputee patient but also the industrial robotic arm.

Project Objectives

The objective of this project is to make robotic arm for trans amputee patients

Project Implementation Method

Methodology starts with the collection of EMG signal from the five male healthy subjects. The analog EMG sensor is used for acquiring signal , the electrodes place at the two muscles “Flexor Digitorum Superficialis” (FDS) and “Extensor Digitorum Communis” (EDC). The electrodes place at the right forearm of the subjects. The goniometer will be tie at the elbow joint for measuring the angles of elbow. The subject will stand and do the five gestures at different positions. After recording the signal the data will be analyzing by following three steps “the pre-processing” , “the feature extraction” and “the classification”.

The whole data analysis will be done by MatLab Software the pre-processing remove the noise and unwanted signal and makes the signal smooth and usable the feature extraction will extract the hidden information of the EMG data. The features of EMG data that will be extracted are mean absolute value (MAV), waveform length (WL), zero crossings (ZC), slope sign changes (SSC), mean, variance,root mean square, standard deviation and mean absolute deviation. After achieving performance by using these features more features will be used for better performance. We have the capacity of using many different features for the best performance of the Artificial Hand.  After feature extraction we will do classification of EMG signal by using different techniques of neural network to train the model. The Multiclass classification will be done by making the sigmoid and softmax function. After the software work the hand assembling and integration of software and hardware will be done by using arduino support package of MatLab.

Benefits of the Project

Benefits to trans amputee patients

Technical Details of Final Deliverable

Emg signals acquire from subject and classify its feature using neural network and then feed it to actuator to move robotic arm

Final Deliverable of the Project

HW/SW integrated system

Type of Industry

Medical

Technologies

Artificial Intelligence(AI), 3D/4D Printing, Robotics

Sustainable Development Goals

No Poverty

Required Resources

If you need this project, please contact me on contact@adikhanofficial.com