ABC Wheelchair

Author: Bryce Beaumont

Beaumont, Bryce, 2020 ABC Wheelchair, Flinders University, College of Science and Engineering

Terms of Use: This electronic version is (or will be) made publicly available by Flinders University in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. You may use this material for uses permitted under the Copyright Act 1968. If you are the owner of any included third party copyright material and/or you believe that any material has been made available without permission of the copyright owner please contact with the details.


Wheelchairs can provide a sense of freedom to people who are unable to walk as many others can. Producing a smarter wheelchair is a necessary step for extending this freedom to individuals who possess motor and/or visual impairments that cause operating a wheelchair difficult or impossible. The ABC Wheelchair at flinders university is pursuing the development of an addon technology for powered wheelchairs that can provide autonomous navigation and allow for multiple input methods for users to utilise.

This project researches the installation of SLAM on the ABC wheelchair and the development of a guided doorway navigator that can guide the chair through narrow doorways. The doorway navigator used a LiDAR to return range information on the environment and recognise the position of door frames by measuring the range differences between data. Assumptions were made when ranking the potential doorways detected to allow for successful detection, such as the navigator being triggered by an external process so doors would already be close to the chair.

Using the LiDAR the system was capable of performing accurate SLAM using the Cartographer system by Google, mapping and localising the position of the chair over time. Furthermore the doorway navigator could successfully distinguish open doorways from LiDAR range data and a velocity controller provided safe travel through the opening.

Additional research areas were focussed on the development of alternative input methods to provide accessible technology to a diverse audience through the means of developing a pupil detector for future gaze trackers and testing the capacity of the MyCroft voice assistant to control the wheelchair.

Pupil detectors built need further development before they can be used for gaze trackers. The first detector is highly accurate, finding pupil location within 1 pixel of error, however lacks responsiveness. The second detector is highly responsive although has higher levels of error.

The MyCroft voice assistant could control chair behaviour and allowed for multiple keyword or phrases to be used as triggers but required network access. A local speech recogniser may be run on the machine to remove the need of network access however is limited by system resources to run the deepnet speech recognition locally, slowing down the execution of real-time systems required for safe autonomous navigation when approaching obstacles.

Keywords: Control Interface, HCI, Smart Wheelchair, ABC Wheelchair, Assistive Technology, Impairment, Disability, Gaze Tracking, Pupil Recognition, LiDAR, Navigation, SLAM, Autonomous Wheelchair

Subject: Engineering thesis

Thesis type: Masters
Completed: 2020
School: College of Science and Engineering
Supervisor: Dr. Nasser Asgari