Pengolahan Sinyal EEG untuk Kendali Transhumeral Robotic Arm Berbasis Brain Computer Interface dan Convolutional Neural Network

Darmawan, Joanna (2025) Pengolahan Sinyal EEG untuk Kendali Transhumeral Robotic Arm Berbasis Brain Computer Interface dan Convolutional Neural Network. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5022211214-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Penyandang disabilitas, khususnya tunadaksa, mengalami kesulitan dalam menggerakkan tubuh dan memerlukan bantuan orang lain atau alat bantu untuk mengatasi keterbatasan motorik. Salah satu teknologi yang dikembangkan untuk membantu adalah Brain Computer Interface (BCI), yaitu sistem yang memungkinkan komunikasi langsung antara otak dan perangkat elektronik eksternal. Salah satu metode BCI yang non-invasif adalah menggunakan sensor untuk merekam sinyal electroencephalogram (EEG) dari otak. Pada Tugas Akhir ini, sinyal EEG direkam menggunakan Headset NeuroSky Mindwave Mobile 2 melalui ThinkGear Communication Protocol dan diolah menggunakan Butterworth Filter serta Short time Fourier Transform untuk diklasifikasikan ke dalam lima pita frekuensi berbeda: gamma, beta, alfa, teta, dan delta. Kelima pita ini menunjukkan perbedaan tingkat konsentrasi yang kemudian diklasifikasikan menggunakan algoritma Convolutional Neural Network (CNN). Model CNN yang dilatih berdasarkan spektrogram kondisi Fokus dan Rileks menunjukkan performa akurat dengan akurasi training 95% dan validation 97%. Model ini diintegrasikan dengan sistem mikrokontroler untuk mengendalikan robot tangan prostetik. Robot memiliki dua mode: mode manual, dimana gerakan yaw, pitch, dan roll dikendalikan oleh gerakan kepala melalui sensor IMU, dan membuka-menutup jari dikontrol oleh sinyal EEG; serta mode otomatis, yang sepenuhnya dikendalikan oleh sinyal otak untuk mengambil gelas, minum, dan mengembalikannya gelas. Hasil pengujian menunjukkan waktu respons rata-rata 4,8 detik untuk Fokus dan 9 detik untuk Rileks. Hasil ini menunjukkan potensi implementasi sistem sebagai alat bantu gerak bagi penyandang disabilitas, khususnya tunadaksa, dalam membantu mereka dalam melakukan aktivitas sehari-hari.
==================================================================================================================================
People with disabilities, especially those with quadriplegia, often face difficulties in moving their bodies and require assistance from others or assistive devicesto overcome motoric limitations in carrying out tasks. One of the technologies developed to help is the Brain Computer Interface (BCI), a system that allows direct communication between the brain and external electronic devices. One of the non-invasive BCI methods is using sensors to record electroencephalogram (EEG) signals from the brain. In this Final Project, EEG signals were recorded using the NeuroSky Mindwave Mobile 2 Headset via the ThinkGear Communication Protocol and processed using the Butterworth Filter and Short-time Fourier Transform to be classified into five different frequency bands: gamma, beta, alpha, theta, and delta. These five bands show differences in concentration levels which are then classified using the Convolutional Neural Network (CNN) algorithm. The CNN model trained based on the Focus and Relax condition spectrograms showed accurate performance with a training accuracy of 95% and a validation accuracy of 97%. This model was integrated with a microcontroller system to control the prosthetic hand robot. The robot has two modes: manual mode, where yaw, pitch, and roll movements are controlled by head movements via the IMU sensor, and opening and closing the fingers is controlled by EEG signals; and automatic mode, which is fully controlled by brain signals to pick up a glass, drink, and return the glass. The test results show an average response time of 4.8 seconds for Focus and 9 seconds for Relax. These results
show a potential for implementing the system as a mobility aid for people with disabilities, especially those with quadriplegia, in helping them in carrying out daily activities.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Electroencephalogram, Brain Computer Interface, Butterworth Filter, Short time Fourier Transform, Convolutional Neural. Electroencephalogram, Brain Computer Interface, Butterworth Filter, Short-time Fourier Transform, Convolutional Neural Network Network
Subjects:	Q Science > QP Physiology > Q376.5 Electroencephalography (EEG) T Technology > TJ Mechanical engineering and machinery > TJ211.4 Robot motion T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5102.9 Signal processing.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Joanna Darmawan
Date Deposited:	23 Jul 2025 01:02
Last Modified:	23 Jul 2025 01:02
URI:	http://repository.its.ac.id/id/eprint/120590

Actions (login required)

View Item