Analysing the Impact of Vibrations on Smart Wheelchair Systems and Users

EasyChair Preprint 7747

Abstract

Mechanical vibrations due to uneven terrains can significantly impact the accuracy of computer vision systems installed on any moving vehicle. In this study, we investigate the impact of mechanical vibrations induced using artificial bumps in a controlled environment on the performance of smart computer vision systems installed on an Electrical powered Wheelchair (EPW). Besides, the impact of the vibrations on the user's health and comfort is quantified using the vertical acceleration of an Inertial Measurement Unit (IMU) sensor according to the ISO standard 2631. The proposed smart computer vision system is a semantic segmentation based on deep learning for pixels classification that provides environmental cues for visually impaired users to facilitate safe and independent navigation. In addition, it provides the EPW user with the estimated distance to objects of interest. Results show that a high level of vibrations can negatively impact the performance of the computer vision system installed on powered wheelchairs. Also, high levels of whole-body vibrations negatively impact the user's health and comfort.

Keyphrases: Mechanical vibration, Powered Wheelchair, Smart Systems, Vibration Impact, Whole-body vibration, computer vision, deep learning, pixel classification, semantic segmentation, semantic segmentation system performance

@booklet{EasyChair:7747,
  author    = {Elhassan Mohamed and Konstantinos Sirlantzis and Gareth Howells},
  title     = {Analysing the Impact of Vibrations on Smart Wheelchair Systems and Users},
  howpublished = {EasyChair Preprint 7747},
  year      = {EasyChair, 2022}}