Applications of Probability Density Function for Instrumented Wheelchair Control Based on EMG Signals from Arm and Forearm Muscles

Mohd Hanafi Muhammad Sidik; Saiful Anwar  Che Ghani; Abdul  Nasir; Norain  Abdullah

doi:10.37934/araset.65.2.1928

Authors

Mohd Hanafi Muhammad Sidik Faculty of Mechanical and Automotive Engineering Technology, Universiti Malaysia Pahang Al-Sultan Abdullah, Pahang, Malaysia
Saiful Anwar Che Ghani Faculty of Mechanical and Automotive Engineering Technology, Universiti Malaysia Pahang Al-Sultan Abdullah, Pahang, Malaysia
Abdul Nasir Faculty of Mechanical and Automotive Engineering Technology, Universiti Malaysia Pahang Al-Sultan Abdullah, Pahang, Malaysia
Norain Abdullah Bio-Coke Research Institute, Kindai University, Osaka, Japan

DOI:

https://doi.org/10.37934/araset.65.2.1928

Keywords:

Electromyography, probability density function, Arduino, instrumented wheelchair, Myoware

Abstract

Patients with muscle weakness or stroke continue to use wheelchairs as essential mobility aids. Some stroke patients who have restricted movements of their hands due to muscle weakness. As an alternative to propel wheelchair more easily, an instrumented wheelchair that has Power Assist System (PAS) with Electromyography (EMG) interface would help the patients by providing additional forces. The device is consisting of Myoware Muscles sensors, Arduino board, motor driver and installed with machine learning Probability Density Functions (PDF) classifier to recognise hand movement pattern to instruct PAS to move forward or stop. 3 participants volunteered in this study to evaluate the classification accuracy of the device by recording EMG signals from Biceps Brachii (BIC), Triceps Brachii (TRI), Flexor Digitorum (FIX) and Extensor Digitorum (EXT) muscles. PDF showed a good response by having average classification accuracy as high as 97.85% and lowest is 89.99%. This finding shows the capability of PDF classifiers in EMG applications.