Photos courtesy of Ryan Mintz
At the age of 21, Ryan Mintz managed to do with a few thousand bucks what the likes of DARPA have spent millions of dollars doing: he built a robotic arm that responds to human brain waves.
Mintz and two classmates at the University of Toronto, Christopher Trudeau and Arturo Iacobelli, built their system using a commercially available Emotiv EPOC headset, a laptop, and a robot arm supplied by the university.
The headset captures electromyographic (EMG) signals—the electrical currents given off by nerves in skeletal muscles—and their homebrew software then processes the signal into a numerical value from zero to five that corresponds with a movement the arm can make. This means that clenching your jaw and winking your eyes in sequence can direct the arm to perform a task like picking up an object.
The trio worked just once a week for a few months and spent under eight hundred dollars, a far cry from the fifty thousand you’d have to drop to obtain a top of the line prosthetic with a similar level of functionality.
“People see anything to do with EEG [electroencephalography] and they immediately go, like, full sci-fi without explaining how it actually works, which leads to a lot of false views,” Mintz told me when I him asked how he felt about having his name in e-lights overnight. EEG involves recording electrical signals in the brain and mapping them to a control output like a robot arm—what often gets dubbed “mind control” in the press.
Though the project was called “EEG-Based Features for Control of a Robotic Arm,” the team ultimately found that the Emotiv headset was too inaccurate make EEG data useful to them. EMG offered a simpler solution. After recording the raw EMG data with the headset, their signal processing software assigns it a value ranging from zero to five that corresponds with an action for the arm. The system is also highly customizable.
“Our software is based on a machine learning approach for recognizing facial movements, which allows it to be trained for an individual user, whereas the Emotiv software is based on a database,” Mintz explained.
The DARPA-funded DEKA “Luke” arm is another futuristic prosthetic that uses EMG. According to an FDA press release, “the EMG electrodes in the DEKA Arm System convert electrical signals into up to 10 powered movements.” Other prosthetics are using the technology as well, but Mintz’s system adds some decidedly cyberpunk flair.
“All you need is any laptop—we didn’t use a high end laptop by any means—and straight out of the box the Emotiv headset worked pretty much fine, and then you just need a Matlab code,” Mintz said. It’s entirely possible to do in your garage, you don’t need any special tools. The arm might be expensive, but on the flip side you can make a very basic robotic arm fairly cheaply.”
As the cost of building DIY robots continues to drop and instructions and code become more readily available online, homemade prosthetics could become cheaper and more commonplace. And the need is there, as demonstrated by Sun Jifa, a Chinese inventor who has built over eight hundred DIY prosthetics for amputees who can’t afford to buy them at normal market price.
Mintz and his teammates have moved on from the project and put their undergraduate years behind them, but they’ve given their hardware, software, and findings to the school in the hopes that next year another group of students will pick up where they left off. They’ve also put their Matlab code online for anyone to download.
Although the trio’s project isn’t likely to become a market competitor for DEKA, they showed that building an advanced prosthetic yourself is entirely possible. “All we wanted to do was advance it a little bit farther and show that EEG is so commonplace now… All you have to do is buy the software and three guys working together once a week can make something out of it.”