Parama Diptya Widayaka, Pradini Puspitaningayu, Sayyidul Aulia Alamsyah, Endryansyah Endryansyah, Lusia Rakhmawati, Paramitha Nerisafitra, Rifqi Abdillah, Akbar Wildhanata, Haikal Alif Eyrlangga
In our daily life, we often find people with certain disabilities or certain physical limitations, such as visually impaired people. People with visual impairment often require assistance from others to perform daily activities. We have seen people with visual impairment using walking sticks to help them walk on the street or pedestrian. With the development of technologies, many studies have been conducted to create disability aids or assistive technology for people with disabilities, especially smart blind sticks with function to detect objects in front of people with limited vision, where the smart blind stick is equipped with a sensor to detect the presence of an obstacle in front of them. However, these aids are still limited to detect the presence of an object by identifying the type of the object. In this study, a smart blind stick system was created by implementing artificial intelligence to detect objects in front of people with visual impairment, where this system applies Tiny-ML that can be implemented on edge devices such as microcontrollers. In the results of this study, the system works well to recognize objects such as cars, motorcycles, road separators, and tactile guides in front of people with visual impairment by an F1 score of 78.1 percent. This object detection also gives a good result from hardware implementation with inferencing time 1368ms, 119.4K peak ram usage, and 90.2K flash usage. © 2025 IEEE.
Universitas Negeri Surabaya, Dept. of Electrical Engineering, Surabaya, Indonesia; Universitas Negeri Surabaya, Dept. of Informatics Engineering, Surabaya, Indonesia
