PROSTHETIC HAND USING EMG SIGNALS

1. Prosthetic Hand: A prosthesis is defined as an artificial device that replace a missing body part lost through trauma, disease, or congenital conditions Prosthetic hands are designed to provide a replacement for upper limb amputees Our hand will be Externally powe

Project Title

PROSTHETIC HAND USING EMG SIGNALS

Project Area of Specialization

Artificial Intelligence

Project Summary

1. Prosthetic Hand:

A prosthesis is defined as an artificial device that replace a missing body part lost through trauma, disease, or congenital conditions Prosthetic hands are designed to provide a replacement for upper limb amputees

Our hand will be Externally powered prosthetic hand. Externally powered prostheses are devices that receive their power from sources other than the user’s body as shown in fig These devices are more functionality than a simple body powered prosthetic device.

2. Selection of sensor:

After a thorough analysis of currently available EMG signal recording electrodes/sensor we select Gravity Analog EMG Sensor, a new device for EMG signal recordings. This sensor has many advantages over the commonly used electrodes. This sensor works on the principle of surface electrodes. 

“our method is non-invasive i.e. we will install sensor on external surface of hand”

3. Machine learning algorithms

We will use machine learning for signals procseesig, and it is the major and main part of our project. 

Learning algorithm is an adaptive method by network computing units self-organizes to realize the target (or desired) behavior. Machine learning is about learning to predict from samples of target behaviors or past observations of data. These algorithms have the ability to learn without being explicitly programmed.

"The algorithm of machine learning that we will used in our project is ANN “Artificial Nueral Network”

Why Artificial neural networks:

ANN can be used to extract patterns and detect trends that are too complex to be noticed by either humans or other computer techniques with their remarkable ability to derive meaning from complicated or imprecise data. The advantages of ANN includes:

• Adaptive learning

• Real time operation

• Fault tolerance

4. SIGNAL AND IMAGE PROCESSING WITH MATLAB/PYTHON ON RASPBERRY-PI PLATFORM:

For signal processing the controller we will use is Raspberry pi, which is super fast in signals processing, so can help for real time signal processing..

SUMMARY OF THE CHAPTER:

We are Actually designing prosthetic hand in which we will take analogue input as EMG signals through “analogue gravity sensor” the signals are then sent to the ADC  which take analogue input and produce digital output, the digital output produced by ADC then send to the controller which is Raspberry Pi, in Raspberry Pi signals are processed through different machine learning algorithms specially neural networks which classify signals for different movements of hand the controller then send processed signals to Servo Motors placed in 3D printed hand. Servo Motor moves according to programmed specification hence a gesture is produced.

Project Objectives

2.1: Project Background

Myoelectric prostheses research involves using surface electroMyography (sEMG), a noninvasive technique that measures muscle activation potentials through the use of electrodes that can detect such sEMG signals through contact with the skin.Prosthetic amputee rehabilitation involves using prosthesis, a man-made device which may replace a missing body part, to recover lost functionality. Research in rehabilitation robotics using Myoelectric prostheses that allow for developing and testing movement recognition algorithms has been performed in the past.

2.2: Main Obejctives:

• To identify different hand poses based in the raw EMG signals using ANN
• To develop a non-invasive control method for prosthetic hand, using EMG signals and machine learning.
• To Control a prosthetic hand in real time
• To control a 3D printed prosthetic hand by hand gestures.
• To develop comparatively a very low cost Prosthetic Hand.
• To establish a deep relation between EMG signals and AI

Our work will focus on the control method instead of 3D design of prosthetic hand. The 3D printed hand will mimic the gestures performed by the user. Since the 3D hand has limited degrees of freedom in such cases we will map the user’s gesture to a different gesture in the prosthetic hand depending upon its degrees of freedom.

2.3: Motivation and need

Pakistan is at the forefront of the war against terrorism since last decades, and as a result, has been severely affected by a rising number of patients with traumatic conditions and disabilities. Both civilians and security forces have been affected. The number of amputees in the last decade has increased. Although there are a few centers in the public and private sectors providing prosthetic services, there is generally a lack of comprehensive multidisciplinary amputee rehabilitation services within the majority of them

2.4: Problem definition

The number of physically handicapped people requiring someone’s assistance in everyday life has been increasing in recent years. These are the people who suffer from the lost of limbs such as arms or legs due to accidents, disease, trauma, congenital defects and so on. They faced a lot of difficulties and inconveniences in their daily life movements and activities. Hence most of them require assistance from their family members. However the quality of life of people with limb deficiency can be improved with the aid of a prosthetic hand.

Project Implementation Method

3.1 Introduction:

The main purpose of project is to identify the hand gestures by pattern recognition of raw EMG data. A machine learning model is used to identify and classify different hand gestures. Artificial neural network(ANN) is chosen as learning model, which is particularly useful for classification purpose

3.2 Control system Architecture

The formal scheme of proposed control strategy is composed of several modules:.

•    EMG Signal Detection
•    MATLAB / PYTHON wrapper
•    Signal pre processing

              Class labeling

              Feature extraction

              Feature normalization

              Principle component analysis(PCA)

•    EMG pattern recognition classifier
•    Prosthetic hand controller
•    Prosthetic Hand

3.3: EMG Signal Detection/recording

A  precise  recording  of  sEMG  signals  is  prerequisite  for  good  performance  of  system. MYO  armband  is  used  in  our  project  instead  of  conventional  electrodes  for  EMG  signal recording.  MANY HUNDRED  readings  are  taken  for  different  hand  gestures  as  shown  i readings are taken in intervals to avoid muscle fatigue, which effects the EMG data. For each reading the data is stored as a comma separated files in the laptop.

3.4: Sensor Armband placement

3D Hand Design:

First we will describe some methods repeated every time we intended to acquire new data, whether it is training or testing data the acquisition protocol will be the same. In every data acquisition session, the Myo armband was placed in a healthy left arm, trying to pay special care to place all the pods in the exact same position every time as shown in fig 3.1

Since all readings are not taken at once, therefore we mark the position, we will try our best to perform these geustures and much more features.

3.5:  MATLAB/PYTHON WRAPPER

MATLAB  wrapper  are  code  bindings  used  for  parsing  EMG  data  from  MYO  sensor   into MATLAB  for  further  processing.  we can parse the data from the spatial and gestural data from the Myo directly to MATLAB for further processing. Myo Armband Manager. This is a compilation  of  codes  thsses  represent  a  physical Myo object hand handles all the data storage and transmission in real time. Our signal will look like this:

3.6: Signal pre processing

EMG signal could not be modeled and processed easily because of their complex, sophisticated structure and mathematical formulation. However, recently the development of machine learning algorithms and their use in pattern recognition gives an opportunity to extract valuable information from bio-signals. class labeling, features extraction, features normalization and principle component analysis(PCA) are performed in EMG pre-processing.

3.7: Class labeling

3.8: Architecture of ANN:

After analysis of different architecture we select a network architecture with one input layer, one hidden and one output layer:

Components of ANN

The main components of ANN model are as follow:

1.     Input of ANN
2.     Output of ANN
3.     parameters/weights of ANN
4.    Biases nodes
5.    Input layer
6.    Hidden layer
7.    Output layer
8.    Neuron
9.   Sigmoid function

Benefits of the Project

1. Ease of motion

The ease of motion is very important, especially when it comes to completing daily tasks. You’ll find that flexing and bending the prosthetic arm will give you great benefits.

2. Ability to hold a certain amount of weight

Chances are, you’ll most likely want to hold some weight using your residual arm. This can be possible when you have a high-quality upper extremity prosthetic.

3. Perform daily tasks with ease and quickness

Depending on the type of prosthetic, hand, and socket used, you can have a fast, powerful prosthetic to meet your daily needs.

4. Comparitively Natural arm movements

When an upper limb prosthetic is fully functioning and designed the right way, you most likely will not experience many restrictions when it comes to natural arm movements. You should definitely choose someone who is experienced with design and function of prosthetics—choose Georgia Prosthetics!

5. Reflexes

With this  quality of  prosthetic system, you can still experience regular reflexes to aid in your daily life. You want a system that is designed to meet your lifestyle, though.

6. Comfortable fit:

althogh it is a an open source research but its design is selected after a long online and offline research.You can experience the advantage of a comfortable fit when the prosthetic  designed and fitted appropriately.

others benefits may b as follow:

• No skin problems
• Stable and safer standing and sitting
• A sense of the artificial limb belonging to the body
• Easy and quick attachment and removal
• Improved quality of life for amputee

Technical Details of Final Deliverable

Our  final deliverable will be 3D printed prosthetic hand in which and which Servo  Motors will be installed for the movement of fingers to perform different gestures the controller which is Raspberry Pi will also be installed inside the 3D hand upon detecting KPMG signals controller will take the signals after processing the signals controller will send the signal the Servo Motors which are installed inside the 3D hand.

our hand will contgain:

1. 3D printed hand
2. Servo motors
3. Raspberry-pi as Controller
4. Emg sensors

Final Deliverable of the Project

HW/SW integrated system

Type of Industry

IT , Others , Health

Technologies

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

Sustainable Development Goals

Good Health and Well-Being for People

Required Resources

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