The goal of this project is to develop a system for hand exoskeletons control in real time. Robotic systems are useful for rehabilitation therapies due to their ability to help patients perform repetitive movements. Hand recovery is especially critical because hands are necessary to perform many daily life activities. Exoskeletons developed for hand rehabilitation can benefit from a real-time control system that activates the robotic devices at the same time as the patient is performing a movement. Real-time control of these systems can be achieved using different methods. In this project, electromyographic (EMG) signals from the patient’s forearm are used. The controlled robotic systems are soft hand exoskeletons actuated with Shape-Memory Alloys (SMA) wires. The SMA wires are controlled with a microcontroller. The main objective of these exoskeletons is to help the patient with the movement of grasping an object and releasing it afterwards. Machine learning is used to detect the intention of the patient to grasp or release an object based on the patient’s EMG signals. Once one of these movements is detected, real-time communication with the microcontroller is achieved and the necessary SMA wires are activated. The system is developed in Matlab, and it involves signal acquisition, signal rectification, signal segmentation, feature extraction, dimensionality reduction, signal classification, and communication with the microcontroller. To differentiate between the movements of grasping and releasing an object, three different classifiers will be tested: Artificial Neural Networks (ANN), Support Vector Machine (SVM) and K-Nearest Neighbors (KNN). Their performance will be compared using a confusion matrix and the best one will be selected for the system’s algorithm. Control is achieved with a time delay of less than 1 second for the action of grasping and of less than 2 seconds for the action of releasing is achieved, almost accomplishing the objective of developing a real-time control system. Several improvements ...
Real-time EMG Control for Hand Exoskeletons
2018-07-04
Theses
Electronic Resource
English
DDC: | 629 |
WEARABLE SENSORS BASED REAL-TIME MONITORING OF GAIT KINEMATICS FOR LOWER EXTREMITY EXOSKELETONS
BASE | 2021
|Wearable robots : biomechatronic exoskeletons
Catalogue medicine | 2008
|Design and Control of Exoskeletons for Planetary Exploration
SAE Technical Papers | 1995
|