About Curriculum Vitae Projects Resources Contact
Eye Controlled Wheelchair
Status: Ongoing as Computer-Wheelchair Interface
This adventure began as my senior design project, for which Andrew Catellier and I received the Best Team Project award. After graduation, I continued the project, and it became the focus of my master's thesis. Due to complications in the filtering and analysis of EOG (electrooculogram) signals, the wheelchair served only as a proof of concept, but could not be tested by a subject with disabilities. I have decided to take a new approach, which begins with designing a universal Computer-Wheelchair Interface.
The prompt was to design a way for a person who could not use his/her hands to control a wheelchair. This is currently accomplished through many methods, including arrays of proximity sensors placed in a headrest, chin/mouth joysticks, "sip and puff" units that sense the pressure of a user's breath in a tube, single switch scanning methods that can utilize a variety of "ability switches," and many more. Unfortunately, the needs and abilities of each user vary greatly, as does the effectiveness of each control method. We decided to add another option to the collective toolbox: eye movement.
Eye tracking is already used by people with disabilities for communication. A person who cannot speak may use an AAC (Augmentative and Alternative Communication) device, the most complicated of which are tablet PCs with specialized software installed. Several such devices are now available with infrared VOG (videooculogram) based eye tracking, including offerings from Tobii, Dynavox, EyeTech, and more. Our approach instead utilizes EOG (electrooculogram), which does not require any hardware to placed in the user's field of view.
While EOG can sense both vertical and horizontal movement of the eyes, we opted to use only horizontal movement for steering, as it seemed impractical for a user to look up and down in order to move forward and backward. Instead, the chair cycles through modes, including "conversation," "pivot," "forward," and "backward." These modes are controlled by a pair of 3.5 mm (1/8") mono jacks, which are compatible with most "ability switches." For testing, a "sip and puff" unit was used.
The system is controlled by a Motorola HCS12 "Minidragon" microcontroller, and powered by four 7.2 volt battery packs (in order isolate the user from the wheelchair's power system).
Unfortunately, the DC drift inherent in EOG signals makes them very difficult to use for control of such a system. The EagleEyes system utilizes a DC blocking filter to circumvent this problem, but while continually drifting toward center is acceptable for a mouse cursor, it is impractical for wheelchair control. In addition, the process of reverse engineering and hacking the electronics of the existing wheelchair is time consuming and introduces liability issues.
I plan to address these problems and take advantage of existing eye tracking technology through a new approach, described in my Computer-Wheelchair Interface project.
For more detail, please see my master's thesis: Expanding Smart Wheelchair Technology for Users with Severe Disabilities.
Copyright © 2012 Gavin Philips. All rights reserved.