Prosthetic Hand Using Surface EMG Signals
Summary 1. Introduction: The goal of this project is to develop a robotic prosthetic arm that can be controlled using EMG signals. This project will consist of three major systems: EMG data acquisition, EMG classification and the robotic pr
2025-06-28 16:34:37 - Adil Khan
Prosthetic Hand Using Surface EMG Signals
Project Area of Specialization Electrical/Electronic EngineeringProject SummarySummary
1. Introduction:
The goal of this project is to develop a robotic prosthetic arm that can be controlled using EMG signals. This project will consist of three major systems: EMG data acquisition, EMG classification and the robotic prosthetic. The EMG signals that are produced while performing a particular hand gesture are never identical, which requires the use of a pattern recognition algorithm. Prosthetic hands can be an important tool to make everyday tasks easier for those who need them. Think about the tasks your hands might do every day, from delicate maneuvers like threading a needle to forceful tasks like hammering a nail. Prosthetic hands vary depending on our goals and needs, but partial hand loss is one of the most common issues addressed by prosthetics. As with many things, recent technological improvements can benefit us if we find ourselves in need of a prosthetic hand. To meet these constraints and to be a practical solution, the system would inevitably have to be developed on an embedded platform.
2. Electrical Components:
The electrical components, which are being used in this project, are;
- Power Supply.
- Servo Motor.
- sEMG-sensor.
- Arduino.
- Wiring.
3. Methodology:
It consists of seven steps.
- Detection of EMG signal.
- The next decision is to determine the sequence of steps (process).
- Noise Reduction.
- Amplification of Signal.
- ADC (For implementing the desired control to the motors)
- Reading of EMG Signal by controller.
- Motor drive and movement mechanism of prosthetic hand.
4. Working:
Input will be taken in the form of EMG signals from subject’s muscles by electrodes of EMG sensor. These EMG signals will be amplified by instrumental amplifier and if there will be a noise in signals due to other signals then it will be removed by subtracting these signals. This signal will be provided to low pass filter, which will remove the high frequency EMG signals. The remaining EMG signals will pass through high pass filter, which will remove the low frequency signals, and only standard frequency signals will remain. These analog signals will be converted to digital via ADC converter. These digital signals will be provided to microcontroller, which will give instructions to run relevant motors, and prosthetic hand will work by the movement of motors and functions end.
5. Flow Chart Diagram:
Objectives:
This project is motivated by the challenge to develop a prosthetic hand that is not heavy, it is very cost-effective (to be affordable by almost every amputee), with effective and efficient mechanisms and capable of reproducing most of a human hand’s movement and gripping patterns. The prosthetic hand is strong enough to hold different object shapes and weight, and its size is comparable to a human hand. As such, the thesis focuses on designing a cost-effective five-finger prosthetic hand. Polylactic acid (PLA) or Nylon is chosen as the material of 3-D printed hand.
Pattern recognition on EMG (Electromyography) signals that are received from human
Muscular movements that are otherwise complex to analyze on some standard methods.
Following are the some of the objectives:
1. Gaining experience and analyzing the stages involved for processing EMG data.
2. Design and Construct a mechanical hand using available tools and materials and control the fingers using servomotor.
3. Perform motor interfacing with Arduino and hand.
4. Use the analyzed EMG signal to control the mechanical hand.
5. Development of a robotic prosthetic arm that can be controlled using EMG signals.
6. Development of user-friendly prosthetic arm.
System Implementation:
1. Data Acquisition:
The synchronized measurements of sensors are shown in every data set. This provides us with EMG data of 64. And the leftmost column is the gesture of the result obtained when results from class 0 to class 3 are registered. So, per line has a structure. Data has been recorded set at 200 Hz, meaning that every line is (no. of sensors)*(1/200) seconds = 40ms record time. A class of gestures (0-3) determines when 64 figures are provided. Rock-0, Paper-1, Spherical-grip2, Alright-3. Movements have been arbitrarily selected and amputees enrolled spontaneously have not faced any physical or mental issues in most cases. They has been embedded via the EMG (Mayo Ware) sensor and evaluated through the surface electrodes. Every activity has been reported with a gap of 6 periods for 20 seconds. During every held organized movement, the tracking system was started. Although while still holding the movement, the tracking system was ceased. Motion takes a minimum of 120 seconds in a static position. They all emerged from the same right forearm within a short period.
2. sEMG Signal Detection:
An accurate exposure of individual action in sEMG is an essential topic in the research of the hand’s motor system. A single-threshold system has been used to expose muscle timing on and off, analyzing the Root Mean Square (RMS) value of the reformed signals to thresholds whose result is based on the mean power of the framework of noise. The frequency range of the sEMG signal is (5-500) Hz. It requires frequency sampling greater than or equal to 1000Hz. The limitations are up to 200 Hz for Myoware Muscle Sensor which has been used in this system. Reliable as well as effective methods such as noise refining, rectification, normalization etc. are needed to process the collected signals accurately as EMG data contains external and removable signals. Additionally, the amplitude ranges for the EMG signal before amplification is 0mV-10mV (5mV). As EMG signal is bipolar so it usually stretches both in positive and negative directions and tends to focus on zero.
3. Support Vector Machine(SVM-KNN):
A difficult situation arises in deciding the kernel function parameters. In an attempt to find a solution for this problems, SVM has been combined with k-nearest neighbor classifier. It is the easy solution for deciding SVM kernel function parameter which reduces difficulty. SVM chooses only one optimized point from the plane where the hybrid algorithm chooses many for a single class. One class can hold many representative data points to represent a class in SVM-KNN. Maximum information gets exploited during the classification process. This system has an accuracy level of 95.398 % which is approximate and higher than other algorithm approaches for classifying EMG data.
Benefits of Prosthetic Hand Using Surface EMG Signals:
Prosthetic device is a device which replaces a missing part of a body by implant an artificial body part. A person who faces the amputation is called amputee. Life of an amputee is not so easy and quite miserable as well. There are lot of technologies developed on the prosthetic hand although they have not much accuracy. The latest technology is EMG based prosthetic hand. It has many benefits over the other technologies and itself as well. Because of these technological advances, there are a great number of benefits to prosthetics.
1. Quick reflexes:
At maximum speed, the prosthetic hand opens about three times as fast as comparable prosthetic hands. With a proportional speed of 300 mm per second, you are even able to react quickly while grasping.
2. Secure hold:
The integrated Auto grasp feature of the Sensor Hand Speed helps you when an object you are holding starts to slip. The gripping force is increased in fractions of a second until the grasped object is in a safe position again.
3. Grasping objects:
Gain freedom of movement in everyday life, at work and in your leisure time. You can grasp and hold various objects, regardless of whether they are light or heavy, small or larger.
4. Replacement:
In addition to accidents, there are many people that suffer from diseases that can lead to amputation. Sometimes, these diseases are hard to fight off and don’t allow a lot of time for the patient to take care of the issue before it is too late. A disease called peripheral arterial disease is one of the most common diseases leading to the amputation process.
5. Process:
Many newly amputees are concerned about how this whole process works. This can be such a big change and create a lot of scared feelings. In the beginning, it is important to set up a consultation just like any other sort of procedure or treatment. An evaluation will help determine which type of design and fit will best suit you. Depending on the area of the limb involved, this will determine which direction the doctor will go for your specific case. This consultation will also be the chance for you to decide the type of movement you can expect after you have received your limb replacement.
6. Advancements:
If you need a replacement in the upper body area, specifically the arm, then the mechanics can become a little trickier to allow you the everyday function that you once had. Some people prefer a robotic limb to help gain control of the prosthetic. With newer advancements, the arm will allow movement to be programmed for a more customized feel. The grip programming is much stronger to allow more access even with a glove covering the hand area.
7. Benefits of Surface EMG:
Surface EMG recordings provide a safe, easy, and noninvasive method that allows objective quantification of the energy of the muscle. It is not necessary to penetrate the skin and record from single motor units to obtain useful and meaningful information regarding muscles.
Technical Details of Final Deliverable:
1. Signal Extraction and Processing:
Electromyography (EMG) has defined as a function of time in terms of amplitude, frequency, and phase. EMG is an electrical current generated in muscles during contraction. The raw EMG signal is a voltage difference measured between recording electrodes. Since the raw EMG signal has positive and negative components, this process can be done using the Rectification which is the translation of the raw EMG signal to a single polarity frequency. Two types of signal’s rectification refer to what happens to the EMG wave when it is processed, these types include full-length frequency and half-length. The full-length frequency was used in our EMG processing that adds the EMG signal below the baseline (usually negative polarity) to the signal above the baseline making a conditioned signal that is Natural and Engineering Sciences 61 all positive, so full-wave rectification takes the absolute value of the signal array of data points
Features measure characteristics from input data, and thus plays an important role when it comes to pattern recognition system design. In this study, feature extraction is carried out on the reference(input) EMG signals to reduce the data dimensionality such that the signal patterns which help to distinguish between the gesture classes were undisturbed. In order to portray an object to be recognized by measurements whose values are alike for objects in same category, and very different for objects in different categories, these characteristics are to be separated. The feature set included MAV (Mean Absolute Value), VAR (Variance), WL (Waveform Length), Kurtosis and Peak.
2. EMG Sensor with Microcontroller:
The microcontroller was used to processes the signals by rectifying and summing. If the magnitude of the relevant sensor exceeds a predefined threshold value; a muscle impulse is said to have been detected on that channel. An EMG sensor is used to amplify, rectify, smooth analog signals. The sensor consists of low-power JFET-input operational amplifiers to turn alternating signal into a direct voltage. Thus, this signal can be read by a microcontroller by using some additional common components of resistors and capacitors. A microcontroller code is written in C++ programming language to convert analog signals into numbers via an internal analog-to digital converter (ADC). This code controls servo motors that are connected to Hand and the microcontroller. An extra servo motor is rotating with a max of 180 degrees. Thus, the hand Natural.
3. Smooth Movement:
Obtain smooth movement of the prosthetic hand.
- Use microcontroller to control individual servos.
- Control multiple servos to obtain basic gestures.
- Smooth out movement by trying different algorithms.
- Test complex gestures.
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 79550 | |||
| Inmoov Right Hand 3d printed | Equipment | 1 | 19800 | 19800 |
| 3d Printed Arm Model | Equipment | 1 | 6000 | 6000 |
| EMG Shield for Arduino | Equipment | 2 | 5500 | 11000 |
| EMG Sensor | Equipment | 2 | 3500 | 7000 |
| Arduino Uno R3 | Equipment | 2 | 900 | 1800 |
| Hitec Standard Economy Servo HS311 | Equipment | 5 | 2950 | 14750 |
| Force Sensitive Resistor (FSR) Sensor | Equipment | 5 | 850 | 4250 |
| Fish Wire 10 meter | Equipment | 2 | 500 | 1000 |
| Disposable Surface EMG Electrodes 1 | Equipment | 1 | 3050 | 3050 |
| AD620 Instrumentation Amplifier | Equipment | 3 | 300 | 900 |
| Invihub Prosthetic Hand 3D Printed DIY kit | Miscellaneous | 1 | 5000 | 5000 |
| Visiting Market & Modeling Project | Miscellaneous | 5 | 1000 | 5000 |