Realtime EMG Signal Processing for the Control of a Motorized Prosthetic Hand

The number of amputees in Pakistan is on the rise due to natural disasters and different acts of terrorism during the last decade. Some other factors include loss of limbs in road accidents, industrial accidents, and birth defects. It is hard for amputees to live a normal life and to perform their d

Project Title

Realtime EMG Signal Processing for the Control of a Motorized Prosthetic Hand

Project Area of Specialization

Artificial Intelligence

Project Summary

The number of amputees in Pakistan is on the rise due to natural disasters and different acts of terrorism during the last decade. Some other factors include loss of limbs in road accidents, industrial accidents, and birth defects. It is hard for amputees to live a normal life and to perform their daily tasks. Apart from this, they are considered incompetent by society. Most of the rehabilitation centers in Pakistan provide passive or 3D printed artificial limbs, but these limbs have very limited functionality. Therefore, it is highly desirable to work on a design and implementation of active and motorized limbs. The major issues are their real-time control and reliability. Electromyography (EMG) provides a solution to control the limbs from the human brain in an intuitive way. So, this project focuses on the real-time processing of EMG signals for control of the motorized robotic hand.

Project Objectives

The goal of this project is the recording and processing of EMG signals for different hand movements.

The hand movements include open-close motion, grasping motion, wrist rotation, and individual finger

movement. These hand movements will be classified in
Real-Time for the control of a motorized prosthetic hand.

The objectives of this project are:

• Literature study regarding the previous work done on EMG signal processing
• Anatomy and Physiological study of the hand and arm
• Selection of appropriate sensors to be used for EMG signal acquisition
• Selection of features and their extraction technique to be implemented in real-time
• Classification of the features of EMG signals for different movements
• Implementation of the Algorithms on Real-Time Operating System
• Generation of Dataset of Recorded EMG Signals
• Demonstration and evaluation of EMG processing

Project Implementation Method

The complete implementation method is broken down to multiple stages. The first is the Recording stage, in which EMG sensors will be used for sensing and recording the EMG signals. After recording, the next stage is
Pre-processing. In this stage, the EMG signal is amplified along with the removal of noise. This amplified signal is then rectified. This rectified signal is digitized and passed through a process to extract its Auto-Regressive (AR) Model parameters. These parameters will serve as features for our feature space. The parameter estimation of the AR modeling is achieved using the Recursive Least Square (RLS) method. Using these estimated parameters, our AR model for a particular hand movement is formed. This completes the Feature Extraction stage.

Next stage is Classification. The Real-Time AR model of the signal is then compared with the stored models for a set of known finger movements and hence desired finger movement is detected. Along with the detection of the finger, the speed of movement is also controlled to obtain the desired angle of the tip of the finger with respect to its base. The operating system for implementing the algorithms will be Real-Time Operating System (RTOS). RTOS introduces the concept of multitasking in Real-Time. The scheduling of tasks and resource management will be done in RTOS. RTOS will make sure that all the deadlines of the system tasks are achieved so that our system processing will be in Real-Time.

Benefits of the Project

The most important benefit of this project is that an amputee will be able to control the prosthetic hand in an intuitive and natural manner. He will have a much wider range of control and will be able to perform his tasks efficiently. It offers many advantages to individuals with an arm or leg amputation compared with a 3D printed passive arm (which fits over the stump of the amputated leg or arm). The attachment of the prosthetic arm will be much more stable and it will provide a wide range of movements, making everyday tasks much easier. A prosthetic arm will not cause pain or skin breakdown when used.

Our proposed techniques can be extended for other limbs e.g. arms, legs, etc. as well. EMG also has other applications in medical fields including diagnosis of muscular diseases and fatigue analysis. EMG signals can also be utilized in
tele-operating systems. All these applications can utilize our signal processing techniques. Hence, overall our research can lay the foundation for other important and useful applications related to EMG.

Technical Details of Final Deliverable

The developed hardware will take inputs from a total of 10 EMG sensors (one EMG sensor pair for each finger to detect the contraction and relaxation states). These signals are processed on Atmel SAM3X8E ARM Cortex-M3 board that will process the signal using Auto-Regressive (AR) modeling to detect different finger movements. The output of the processor is control signals for five different motors. These motors control each finger movement. The system will be capable of moving the finger individually at different speeds and different angles. Each finger is being controlled by a single motor and finger joints are coupled to give a specific trajectory of the tip of each finger.

Final Deliverable of the Project

HW/SW integrated system

Type of Industry

Medical

Technologies

Artificial Intelligence(AI), Robotics

Sustainable Development Goals

Good Health and Well-Being for People

Required Resources

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