Priyanpadma, Sulthon and Prof. Dr.Eng. Fitri Utaminingrum, S.T., M.T (2024) Kendali Kursi Roda Pintar Berdasarkan Suara Pada Lingkungan Bising Menggunakan GFCC dan ResNet50 Berbasis TX2. Sarjana thesis, Universitas Brawijaya.
Abstract
Pada tahun 2023, sebanyak 22,97 juta jiwa atau sekitar 8,5% dari jumlah penduduk di Indonesia merupakan penyandang disabilitas (Kementerian Koordinator Bidang Pembangunan Manusia dan Kebudayaan Republik Indonesia, 2023). Penyandang disabilitas dibagi menjadi beberapa jenis, salah satunya adalah penyandang disabilitas tunadaksa. Penyandang disabilitas tunadaksa memiliki keterbatasan pada fungsi fisik yang akan menjadi kendala dalam melakukan aktifitas sehari-hari. Salah satu alat yang dapat membantu penyandang disabilitas dalam beraktifitas adalah kursi roda. Sementara, penyandang disabilitas tunadaksa yang memiliki kelumpuhan pada bagian tangan maupun kaki akan kesulitan dalam menggerakkan kursi roda secara mandiri. Hal ini disebabkan kursi roda yang ada dipasaran hanya memfasilitasi penyandang disabilitas yang dapat mengendalikan kursi rodanya secara mandiri menggunakan tangan. Maka dibutuhkan kursi roda yang dapat dikendalikan tanpa menggunakan tangan, salah satu solusinya memanfaatkan perintah suara untuk mengendalikan kursi roda secara mandiri. Oleh sebab itu, penulis mengusulkan sistem kendali kursi roda pintar menggunakan perintah suara. Selain itu, agar meningkatkan mobilitas pengguna kursi roda maka perintah suara akan melakukan klasifikasi pada lingkungan bising menggunakan fitur ekstraksi Gammatone-Frequency Cepstral Coefficients (GFCC) dan klasifikasi ResNet50. Adapun hasil akurasi integrasi dilakukan oleh 3 subjek dilingkungan hening dan 3 subjek dilingkungan bising. Hasil yang diperoleh terhadap pengujian lima kelas yaitu maju, mundur, kanan, kiri, berhenti menunjukkan bahwa pengujian pada lingkungan hening memperoleh rata-rata akurasi 88% dan pada lingkungan bising memperoleh rata-rata akurasi 80%.
English Abstract
In 2023, a total of 22.97 million people, or approximately 8.5% of Indonesia's population, were persons with disabilities (Ministry of Human Development and Culture of the Republic of Indonesia, 2023). Persons with disabilities are categorized into several types, one of which is people with physical disabilities. Individuals with physical disabilities face constraints in their daily activities due to limitations in their physical functions. One of the devices that can assist persons with disabilities in their activities is a wheelchair. However, individuals with physical disabilities who have paralysis in their hands or legs will struggle to maneuver a wheelchair independently. This is because conventional wheelchairs available in the market are designed to facilitate persons with disabilities who can control the wheelchair manually using their hands. Therefore, there is a need for a wheelchair that can be controlled without the use of hands, and one possible solution is to utilize voice commands to control the wheelchair independently. Hence, the author proposes a smart wheelchair control system using voice commands. Additionally, to enhance the mobility of wheelchair users, the voice commands will classify the noisy environment using the Gammatone Frequency Cepstral Coefficients (GFCC) extraction feature and ResNet50 classification. The accuracy results were obtained from testing conducted with 3 subjects in a quiet environment and 3 subjects in a noisy environment. The results obtained from testing five classes—forward, backward, right, left, stop—showed that testing in a quiet environment achieved an average accuracy of 88%, while in a noisy environment, the average accuracy obtained was 80%.
Item Type: | Thesis (Sarjana) |
---|---|
Identification Number: | 052415 |
Uncontrolled Keywords: | Disabilitas Tunadaksa, Kursi Roda, Klasifikasi Suara, ResNet50, Gammatone Frequency Cepstral Coefficients, Multiple Disabilities, Wheelchair, ResNet50, Audio Classification, Gammatone Frequency Cepstral Coefficients |
Divisions: | Fakultas Ilmu Komputer > Teknik Informatika |
Depositing User: | Annisti Nurul F |
Date Deposited: | 12 Feb 2024 01:20 |
Last Modified: | 12 Feb 2024 01:20 |
URI: | http://repository.ub.ac.id/id/eprint/215886 |
Text (DALAM MASA EMBARGO)
Sulthon Priyanpadma.pdf Restricted to Registered users only Download (5MB) |
Actions (login required)
View Item |