Deteksi Tangga Turun, Tangga Naik, dan Lantai menggunakan Gray Level Co-Occurrence Matrix dan K-Nearest Neighbors berbasis Raspberry Pi

Ester Nadya Fiorentina Lumban Gaol (Fakultas Ilmu Komputer, Universitas Brawijaya)
Fitri Utaminingrum (Fakultas Ilmu Komputer, Universitas Brawijaya)
Agung Setia Budi (Fakultas Ilmu Komputer, Universitas Brawijaya)

Jurnal Pengembangan Teknologi Informasi dan Ilmu Komputer
Vol 5 No 2 (2021): Februari 2021

A smart wheelchair is a mobility aid that is easy to operate and aids in its users' independence. In its development, the smart wheelchair design not only pays attention to the aspect of convenience but also in terms of safety. Any obstacles that include accessibility obstacles in the environment must be detectable by a smart wheelchair to increase safety. One of these accessibility obstacles is the sudden change in the level of surfaces such as stairs. This research aims to create a system capable of detecting stairs descent, stairs ascent, and floors using the Gray Level Co-occurrence Matrix (GLCM) method to obtain the extraction of characteristic features from stairs descent, stairs ascent, and floors. Then applies the K-Nearest Neighbors (K-NN) classification method to predict what states are detected. Using the GLCM methods with 4 feature (contrast, dissimilarity energy, and homogeneity), distance (d) = 4, q=90° and K-NN with k=4, the system obtained an average accuracy of 93.33% and an average computation time of 1551ms.