The students at McGill NeuroTech challenge themselves to do big things.
Like build a brain-controlled wheelchair in 30 days as an extracurricular activity.
Say what?
That’s the ambitious goal and tight deadline the student group faced this winter to submit their project to the NeuroTechX Student Clubs Competition. (Which they won by the way, but more on that later.)
The wheelchair operates via EEG signal – a user has four electrodes attached to their head – and uses imagined movement.
“It’s almost science fiction-y,” says Raffi Hotter, a first year honours math and computer science student and Schulich Leader scholar who led the software team.
“The person who’s using the wheelchair will just imagine moving to the left and the wheelchair will take them to the left. Or imagine moving to the right, the wheelchair will take them to the right,” Hotter explains.
“They imagine these motions and the wheelchair takes them in those directions. That’s the simple stuff.”
The students also added self-driving features, enabling the wheelchair to move by itself and go around small obstacles.
Getting the wheelchair to do all that required teamwork, self-teaching, and a ton of work.
Marley Xiong and fellow McGill student Jenisha Patel kicked things off by starting to assemble a student team. Among the ideas proposed, the wheelchair was the runaway choice.
Nearly three dozen students worked on the wheelchair they dubbed Milo, for mind-controlled locomotive. They split into teams – data collection, signal processing, machine learning, software and hardware – each saddled with a key piece of the puzzle. They also had a research team.
“We created a pipeline starting with the brain and ending with the wheelchair itself,” Xiong explains in the polished video the students created for the online NeuroTechX competition in February.
“The whole reason for the data collection is that we had to train our algorithms to properly identify what imagining moving to the left means and what imagining moving to the right means, what stopping means,” Hotter says.
The students took an old wheelchair and made it functional again, adding different sensors and a footrest.
There were hiccups along the way, including the technical challenges posed by their inexpensive EEG system with low fidelity.
How much time was involved in creating Milo in 30 days?
“A lot of sleepless nights,” laughs Hotter.
“This was during school. So we were working all the time on it. Many, many hours.”
The effort involved Arts, Engineering and Science students. McGill NeuroTech received funding from the Dean’s Fund in the Faculty of Science that is supported by annual giving to McGill, as well as funding from the other faculties involved. They worked out of Building 21, a research and community space at McGill.
McGill NeuroTech describes itself as a “group of students that builds devices that interface with the brain. We live at the intersection of neuroscience, computer science, engineering, and the human imagination.” Every year, they develop a brain-controlled application.
Their wheelchair won first place – and a $1,000 prize – at the NeuroTechX Student Clubs Competition, beating teams from the University of Toronto, UCLA, and Polytechnique Montréal, among others.
Their feat attracted media attention, and this spring the students presented Milo at Google DeepMind’s office in Montreal.
The students are exploring how to improve the wheelchair and hope to work with patients to test it.
McGill NeuroTech recently launched a gofundme campaign to raise money for materials for new projects and events to attract students to neurotech.
Hotter says the group will either continue with the wheelchair for next year’s project or do something completely different. They also need to find a new space.
Were people surprised that a group of students could build something like this?
“Depends on who you ask, I guess. For a lot of people it’s really surprising,” Hotter says. “This is usually work done by much older people with much more experience. We were teaching ourselves.…Many people hadn’t worked with these signal-processing techniques. There’s a whole course at McGill to teach signal processing. We had to learn it in a couple of days.”
Projects are one of the best ways to learn – “Certainly this one was,” Hotter says, noting the thrill and high motivation level.
“I learned so much from all the other people on the team. People from all different backgrounds. It was really amazing.”