Exosuits — wearable machines that increase strength and endurance — are a flourishing market expected to reduce strain and weakness caused by physical labor, disorders, and even just old age.
Harvard University is developing of one of the most cutting-edge iterations of this technology, a soft exosuit. Think of it as mechanical Under Armour that helps you run, jog, or walk with minimal effort. To take this a step further, researchers at the university have embedded artificial intelligence inside their suits to make sure the wearer and machine are in sync.
Ye Ding, a postdoctoral fellow at Harvard’s School of Engineering and Applied Sciences, believes this type of A.I. will be pivotal in making exosuits commercials viable in the future.
“Every individual needs assistance that is specifically fit for them, so this type of optimization is the right way to go in terms of wearable devices,” Ding, the co-first author of the researcher that melded A.I. and exosuits, tells Inverse. “If you’re trying to get the best performance out of exosuits you have to have something that is figuring out how you should tailor your device at an individual level.”
In a paper published in the journal Science Robotics, Ding and his colleagues explain how they used real-time measurements of human physiological signals, like breathing rate, to tweak the parameters of one of their exosuits.
Their machine learning algorithm calculated the best level of assistance to deliver at any given moment and adjusted the suit accordingly. The study only made use of a “hip extension assistive device,” so it limited the A.I. to only one joint. This first step in this technological combination was a huge success.
“There were huge improvements in the operation of the suit using this technology compared to other state-of-the-art devices and our pervious research,” says Ding. “We saw an improvement of metabolic reduction [reduced energy used] more than 60% which are some promising results about using this type technology in the future”
Before these advancements exosuits used what Ding calls “fixed strategies,” which is essential a default setting for everyone who uses the suit. This would often time use biological joint pattern data to determine the amount of assistance to provide, but a simplified assistance pattern will never get every individual’s profiles right.
Machine learning will make it so exosuits can fit the needs on anyone who uses them, whether it be an elderly couple going on a leisurely stroll to a firefighter sprinting into a burning building.
The human body is constantly changing and the only feasible way, at least the moment, of translating this to robotics is through the lens of A.I.