Intelligent Wheelchair Control System based on Finger Pose Recognition

Iswahyudi Yudi, Khairul Anam, Azmi Saleh


In the old day, wheelchairs are moved manually by using hand or with the assistance of someone else. Users of this wheelchair get tired quickly if they have to walk long distances. The electric wheelchair emerged as a form of innovation and development for the manual wheelchair. This paper presented the control system of the electric wheelchair based on finger poses using the Convolutional Neural Network (CNN). The camera is used to take pictures of five-finger poses. Images are selected only in certain sections using Region of Interest (ROI). The five-finger poses represent the movement of the electric wheelchair to stop, right, left, forward, and backward. The experimental results indicated that the accuracy of the finger pose detection is about 93.6%. Therefore, the control system using CNN can be a potential solution for an electric wheelchair.


Object Recognition; Region of Interest (ROI); Convolutional Neural Network (CNN

Full Text: PDF


  • There are currently no refbacks.