Real-Time Hand-Gesture Recognition for the Control of Wheelchair
Md Rafiul Huda, Md. Liakot Ali, Muhammad Sheikh Sadi · 2022
Currently, there are more than 1 billion disabled persons, who experience issues with the structure or function of their body, or challenges performing a task or action, worldwide. Many of them need to use wheelchairs to go around daily, yet they have been encountering many problems using traditional wheelchairs. This paper develops a prototype of an advanced algorithm for gesture recognition and decision-making for a control system for smart wheelchair (built with advanced and automated features) users, in line with contemporary technologies. Overall, hand and finger tracking using the MediaPipe framework and gesture recognition with decision-making by a mathematical model based on distances among the landmarks and thresholds to control a smart wheelchair are the main components of the proposed system. This method places a greater emphasis on users’ flexibility by requiring less hand movement. The experimental study shows that the proposed method outperforms existing methods.