Thursday, June 4, 2015

A wheelchair controlled by eye movements

Myrijam Stoetzer, 14, and Paul Foltin, 15, are from Duisburg in Germany. They have built a system for controlling a wheelchair using eye movements.

A standard webcam with the infrared filter removed tracks eye movements, and the eye is illuminated by (invisible) infrared light from LEDs to allow the system to work in low light conditions. A Raspberry Pi is used to process the video stream to obtain the position of the pupil and compare it with adjustable preset values representing forward, reverse, left and right. The command is then verified with a switch that is currently manual but that should eventually detect small movements of the tongue or cheek.

Myrijam and Paul

Myrijam and Paul demonstrate their wheelchair control system
Photo credit: basf.de

An Arduino controls recycled windscreen wiper motors via relays to turn custom 3D-printed wheels that sit against the tyres of the wheelchair and push the left and right wheels backwards or forwards to control movement and direction. The camera casing is also 3D-printed to Paul and Myrijam’s own design. The latest feature addition is a collision detection system using IR proximity sensors to detect obstacles.

This is their first project after moving on from LEGO Mindstorms, and they’ve chosen to use Python with the OpenCV image processing library for their first build using a full programming language, teaching themselves as they go along.

After finding that the Raspberry Pi 1 was a little slow to handle the image processing, Paul and Myrijam tried alternatives before switching to the Raspberry Pi 2 when it became available. It’s their preferred option considering cost and speed together, and they have said they’d welcome help to better their system’s performance by improving their code or by using the Pi 2’s four cores more efficiently.

Myrijam and Paul have been competing in Jugend forscht, the German science competition for young people, refining and extending their system as they have progressed through the competition, moving on from a model robotic platform to a real second-hand wheelchair, and using prize money from earlier rounds to fund improvements for later stages. At the weekend, they took their project to the final, where they were judged national winners in the “world of work” category. We’d like to offer them our heartfelt congratulations – we’re monumentally impressed by their work!

You can read more on Myrijam’s blog and on Hackaday, where you can also get in touch with this talented duo if you’re interested in helping them with their excellent project.

The post A wheelchair controlled by eye movements appeared first on Raspberry Pi.

No comments:

Post a Comment