Sistem Navigasi Kursi Roda Elektrik untuk Pasien Penyandang Cacat Fisik Menggunakan Metode Convolutional Neural Network

Sutikno Sutikno (Pascasarjana Universitas Jember Teknik Elektro Universitas Muhammadiyah Jember)
Khairul Anam (Teknik Elektro Universitas Jember)
Azmi Shaleh (Teknik Elektro Universitas Jember)

Jurnal Rekayasa elektrika
Vol 17, No 1 (2021)

Patients with physical disabilities, such as losing a leg or experiencing paralysis, will have difficulty moving from one place to another. As a result, they need someone or a device that can help them to move. One that is often used by patients is a wheelchair. This study proposes an electric wheelchair navigation system that can be controlled by voice commands using the Convolutional Neural Network (CNN) method. CNN is used as the main method for identifying commands embedded on the Raspberry Pi microcontroller. The recorded voice data is then converted to spectrogram images before being fed to CNN. This method is proven to be better in voice command recognition with an accuracy of above 90%. There are five different voice commands: forward, backward, left, right, and stop. Preliminary experimental results indicate that the electric wheelchair is able to move according to the commands given.