Development of a new robust hybrid automata algorithm based on surface electromyography (SEMG) signal for instrumented wheelchair control

Instrumented wheelchair operates based on surface electromyography (sEMG) is one of alternative to assist impairment person for mobility. SEMG is chosen due to good in accuracy and easier preparation to place the electrodes. Motor neuron transmit electrical potential to muscle fibre to perform isome...

Full description

Saved in:
Bibliographic Details
Main Author: Mohd Hanafi, Muhammad Sidik
Format: Thesis
Language:English
Published: 2020
Subjects:
Online Access:http://umpir.ump.edu.my/id/eprint/34436/1/Development%20of%20a%20new%20robust%20hybrid%20automata.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
id my-ump-ir.34436
record_format uketd_dc
institution Universiti Malaysia Pahang Al-Sultan Abdullah
collection UMPSA Institutional Repository
language English
topic TJ Mechanical engineering and machinery
spellingShingle TJ Mechanical engineering and machinery
Mohd Hanafi, Muhammad Sidik
Development of a new robust hybrid automata algorithm based on surface electromyography (SEMG) signal for instrumented wheelchair control
description Instrumented wheelchair operates based on surface electromyography (sEMG) is one of alternative to assist impairment person for mobility. SEMG is chosen due to good in accuracy and easier preparation to place the electrodes. Motor neuron transmit electrical potential to muscle fibre to perform isometric, concentric or eccentric contraction. These electrical changes that is called Motor Unit Action Potential (MUAP) can be acquired and amplified by electrodes located on targeted muscles changes can be recorded and analysed using sEMG devices. But, sEMG device cost up to USD 2,100 for a sEMG data acquisition device that available on market is one of the drawback to be used by impairment person that most of them has financial problem due to unable to work like before. In addition, it is a closed source system that cannot be modified to improve the accuracy and adding more features. Open source system such as Arduino has limitation of specifications that makes able to apply nonpattern recognition control methods which is simpler and easier compared to pattern recognition. However, classification accuracy is lower than pattern recognition and it cannot be applied to higher number participants from different background and gender. This research aims are to develop an open-source Arduino based sEMG data acquisition device by formulating hybrid automata algorithm to differentiate MUAP activity during wheelchair propulsion. Addition of hybrid automata algorithm to run pattern and non-pattern recognition based control methods is an advantage to increase accuracy in differentiating forward stroke or hand return activity. Electrodes are placed on Biceps (BIC), Triceps (TRI), Extensor (EXT), Flexor (FIX) and MUAP activity recorded for 30 healthy persons. Then, experiment result was validated with simulation result using OpenSim biomedical modelling software. Mean, standard deviation (SD), confidence interval (CI) and maximum point different (MPD) of MUAP were calculated and to be used as thresholds for non-pattern recognition control method in method selection experiment. Meanwhile, pattern recognition is using Probability Density Function (PDF) to determine MUAP according to type of activities. Total of ten control methods determined from population and individual data were tested against another 10 healthy persons to evaluate the algorithm performance. Assessment of each control method done by misclassification matrix looking at True Positive (TP) and False Negative (FN) of power assist system activation period. Developed sEMG data acquisition device that is operated by Arduino MEGA 2560 and Myoware muscle sensors with sampling rate of above 400Hz successfully recorded MUAP from four arm muscles. Furthermore, 2.5 ms of average data latency for device to record, analyse, validate and creating commands to activate the power assist system. Data obtained from the device shows that most active muscle during wheelchair propulsion is TRI, followed by BIC and matched to OpenSim simulation result. In method selection experiment, 96.28% of average accuracy was achieved and different control methods were selected by misclassification matrix for each of persons. This method would be a control method to activate power assist system and selected based on conditions set in the algorithm. These findings indicated that open source Arduino board is capable of running real time pattern, non-pattern recognition based control methods by producing classification accuracy up to 99.48% even though it is known as just a microcontroller that has limitation to run complex classifiers. At the same time, a device that cost less than USD200 has 400Hz of sampling rate is as good as closed source device that is come with expensive price tag to own it. Based on algorithm evaluation, it shows that one control method couldn’t fit to all persons as per proven in method selection experiment. Different person has different control method that suit them the most. Lastly, BIC and TRI can be reference muscles to activate assistive device in instrumented wheelchair that is using propulsion as indication.
format Thesis
qualification_name Doctor of Philosophy (PhD.)
qualification_level Doctorate
author Mohd Hanafi, Muhammad Sidik
author_facet Mohd Hanafi, Muhammad Sidik
author_sort Mohd Hanafi, Muhammad Sidik
title Development of a new robust hybrid automata algorithm based on surface electromyography (SEMG) signal for instrumented wheelchair control
title_short Development of a new robust hybrid automata algorithm based on surface electromyography (SEMG) signal for instrumented wheelchair control
title_full Development of a new robust hybrid automata algorithm based on surface electromyography (SEMG) signal for instrumented wheelchair control
title_fullStr Development of a new robust hybrid automata algorithm based on surface electromyography (SEMG) signal for instrumented wheelchair control
title_full_unstemmed Development of a new robust hybrid automata algorithm based on surface electromyography (SEMG) signal for instrumented wheelchair control
title_sort development of a new robust hybrid automata algorithm based on surface electromyography (semg) signal for instrumented wheelchair control
granting_institution Universiti Malaysia Pahang
granting_department Faculty of Mechanical and Automotive Engineering Technology
publishDate 2020
url http://umpir.ump.edu.my/id/eprint/34436/1/Development%20of%20a%20new%20robust%20hybrid%20automata.pdf
_version_ 1783732190942593024
spelling my-ump-ir.344362022-06-16T06:25:18Z Development of a new robust hybrid automata algorithm based on surface electromyography (SEMG) signal for instrumented wheelchair control 2020-10 Mohd Hanafi, Muhammad Sidik TJ Mechanical engineering and machinery Instrumented wheelchair operates based on surface electromyography (sEMG) is one of alternative to assist impairment person for mobility. SEMG is chosen due to good in accuracy and easier preparation to place the electrodes. Motor neuron transmit electrical potential to muscle fibre to perform isometric, concentric or eccentric contraction. These electrical changes that is called Motor Unit Action Potential (MUAP) can be acquired and amplified by electrodes located on targeted muscles changes can be recorded and analysed using sEMG devices. But, sEMG device cost up to USD 2,100 for a sEMG data acquisition device that available on market is one of the drawback to be used by impairment person that most of them has financial problem due to unable to work like before. In addition, it is a closed source system that cannot be modified to improve the accuracy and adding more features. Open source system such as Arduino has limitation of specifications that makes able to apply nonpattern recognition control methods which is simpler and easier compared to pattern recognition. However, classification accuracy is lower than pattern recognition and it cannot be applied to higher number participants from different background and gender. This research aims are to develop an open-source Arduino based sEMG data acquisition device by formulating hybrid automata algorithm to differentiate MUAP activity during wheelchair propulsion. Addition of hybrid automata algorithm to run pattern and non-pattern recognition based control methods is an advantage to increase accuracy in differentiating forward stroke or hand return activity. Electrodes are placed on Biceps (BIC), Triceps (TRI), Extensor (EXT), Flexor (FIX) and MUAP activity recorded for 30 healthy persons. Then, experiment result was validated with simulation result using OpenSim biomedical modelling software. Mean, standard deviation (SD), confidence interval (CI) and maximum point different (MPD) of MUAP were calculated and to be used as thresholds for non-pattern recognition control method in method selection experiment. Meanwhile, pattern recognition is using Probability Density Function (PDF) to determine MUAP according to type of activities. Total of ten control methods determined from population and individual data were tested against another 10 healthy persons to evaluate the algorithm performance. Assessment of each control method done by misclassification matrix looking at True Positive (TP) and False Negative (FN) of power assist system activation period. Developed sEMG data acquisition device that is operated by Arduino MEGA 2560 and Myoware muscle sensors with sampling rate of above 400Hz successfully recorded MUAP from four arm muscles. Furthermore, 2.5 ms of average data latency for device to record, analyse, validate and creating commands to activate the power assist system. Data obtained from the device shows that most active muscle during wheelchair propulsion is TRI, followed by BIC and matched to OpenSim simulation result. In method selection experiment, 96.28% of average accuracy was achieved and different control methods were selected by misclassification matrix for each of persons. This method would be a control method to activate power assist system and selected based on conditions set in the algorithm. These findings indicated that open source Arduino board is capable of running real time pattern, non-pattern recognition based control methods by producing classification accuracy up to 99.48% even though it is known as just a microcontroller that has limitation to run complex classifiers. At the same time, a device that cost less than USD200 has 400Hz of sampling rate is as good as closed source device that is come with expensive price tag to own it. Based on algorithm evaluation, it shows that one control method couldn’t fit to all persons as per proven in method selection experiment. Different person has different control method that suit them the most. Lastly, BIC and TRI can be reference muscles to activate assistive device in instrumented wheelchair that is using propulsion as indication. 2020-10 Thesis http://umpir.ump.edu.my/id/eprint/34436/ http://umpir.ump.edu.my/id/eprint/34436/1/Development%20of%20a%20new%20robust%20hybrid%20automata.pdf pdf en public phd doctoral Universiti Malaysia Pahang Faculty of Mechanical and Automotive Engineering Technology