Electromyography gesture classification using CNN-RNN neural network for controlling quadcopters = klasifikasi gestur elektromiografi menggunakan jaringan saraf tiruan CNN-RNN untuk mengendalikan quadcopter

Hakim, Ray Antonius (2019) Electromyography gesture classification using CNN-RNN neural network for controlling quadcopters = klasifikasi gestur elektromiografi menggunakan jaringan saraf tiruan CNN-RNN untuk mengendalikan quadcopter. Bachelor thesis, Universitas Pelita Harapan.

[img] Text (Title)
Title.pdf
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (4MB)
[img] Text (Abstract)
Abstract.pdf
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (132kB)
[img] Text (ToC)
ToC.pdf
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (181kB)
[img] Text (Chapter 1)
Chapter 1.pdf
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (137kB)
[img] Text (Chapter 2)
Chapter 2.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (666kB) | Request a copy
[img] Text (Chapter 3)
Chapter 3.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (786kB) | Request a copy
[img] Text (Chapter 4)
Chapter 4.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (1MB) | Request a copy
[img] Text (Chapter 5)
Chapter 5.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (130kB) | Request a copy
[img] Text (Bibliography)
Bibliography.pdf
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (145kB)
[img] Text (Appendices)
Appendices.pdf
Restricted to Repository staff only
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (253kB) | Request a copy

Abstract

The advancements of technology have improved a lot as of late. One of the sectors currently growing is drone technology. Unfortunately, the controllers used for drones are very traditional (Radio Controllers). It takes time to learn how to effectively control drones. Bringing a big RC controller is also arduous and inefficient. There are many information from the human body that can be extracted by a controller to control a device or object. Electromyography (EMG) signals is something interesting to consider. The problem arises in the method of processing the raw EMG data into accurate gesture recognitions, hence this research proposes a CNN-RNN Neural Network Model approach to identify the gestures. Implementation was made by using Myo, an 8-channel EMG data acquisition device. Myo-Python was used to communicate with the Myo Device. 14000 datasets of 9 different gestures were collected and used to train the CNN-RNN model using 225 epochs. Trained model is then tested on a drone using dronekit to translate the gestures to drone commands and send them from Python to the drone. The results show that a CNN-RNN approach was very effective in classifying gestures from raw muscle data, resulting in an average of 96.6% positive classification for each gesture. The lack of motion tracking limits the amount for the variety of gestures. The classified gesture shows an effective drone control from the simulated drone. Kemajuan teknologi telah meningkat pesat akhir-akhir ini. Salah satu sektor yang saat ini berkembang adalah teknologi drone. Sayangnya, pengendali yang digunakan untuk drone sangat tradisional (Kontroler Radio). Dibutuhkan waktu untuk belajar mengendalikan drone secara efektif. Membawa kontroler radio yang besar juga cukup sukar dan kurang efisien. Ada banyak informasi dari tubuh manusia yang dapat di ekstrak oleh kontroler untuk mengendalikan sebuah perangkat atau objek. Sinyal elektromiografi (EMG) adalah sesuatu yang menarik untuk dipertimbangkan. Masalah muncul dalam metode pengolahan data mentah EMG yang bertujuan untuk dijadikan pengenalan gestur yang akurat, oleh karena itu penelitian ini mengusulkan pendekatan saringan jaraf tiruan model CNN-RNN untuk mengidentifikasi gestur. Implementasi dilakukan dengan menggunakan Myo, perangkat akuisisi data EMG 8-channel. Myo-Python digunakan untuk berkomunikasi dengan Perangkat Myo. 14000 kumpulan data dari 9 gestur berbeda dikumpulkan dan digunakan untuk melatih model CNN-RNN menggunakan 225 iterasi. Model yang terlatih kemudian diuji pada drone menggunakan dronekit untuk menerjemahkan gestur ke perintah drone dan mengirimkannya dari Python ke drone. Hasil penelitian menunjukkan bahwa pendekatan CNN-RNN sangat efektif dalam mengklasifikasikan gestur dari data mentah otoh, menghasilkan rata-rata 96.6% klasifikasi positif untuk setiap gestur. Kurangnya pelacakan motion gestur membatasi jumlah untuk berbagai gestur. Gestur terklasifikasi menunjukkan kontrol drone yang efektif dari drone yang disimulasikan.

Item Type: Thesis (Bachelor)
Creators:
CreatorsNIMEmail
Hakim, Ray AntoniusNIM00000021587rayantonius@gmail.com
Contributors:
ContributionContributorsNIDN/NIDKEmail
Thesis advisorTjahyadi, HendraNIDN0410076901UNSPECIFIED
Uncontrolled Keywords: artificial intelligence; neural network; electromyography; drone; human computer interaction
Subjects: T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7885-7895 Computer engineering. Computer hardware
Divisions: University Subject > Historic > Faculty/School > Computer System Engineering
Historic > Faculty/School > Computer System Engineering
Depositing User: Users 2912 not found.
Date Deposited: 20 Nov 2019 02:36
Last Modified: 20 Apr 2020 07:37
URI: http://repository.uph.edu/id/eprint/5701

Actions (login required)

View Item View Item