The Harvard Biodesign Lab is building in a new level of rapid personalization to a soft exosuit designed to provide hip extension assistance — and that’s just the beginning.
Because every human moves a bit differently — constantly tweaking their movements in order to conserve energy, or metabolic cost — tailoring robotic parameters for an individual user tends to be time-consuming and inefficient. But researchers from the Harvard John A. Paulson School of Engineering and Applied and Sciences (SEAS) and the Wyss Institute for Biologically Inspired Engineering have developed an efficient machine learning algorithm to cut through that variability. The algorithm can rapidly identify the control parameters that work best for minimizing the metabolic cost of walking.
"Before, if you had three different users walking with assistive devices, you would need three different assistance strategies," explains Myunghee Kim, a postdoctoral research fellow at SEAS. "Finding the right control parameters for each wearer used to be a difficult, step-by-step process."
The researchers used “human-in-the-loop” optimization — real-time measurements of physiological signals, such as breathing rate — to adjust the control parameters of the device. As the algorithm honed in on the optimal parameters, it directed the exosuit on when and where to deliver its assistive force.
Results were promising: The algorithm-and-suit combination delivered a 17.4 percent metabolic cost reduction over walking without the device. Perhaps even more impressive was the time frame in which the system was able to automatically learn how to work synergistically with the wearer: around 20 minutes, according to Conor Walsh, an associate professor of engineering and applied sciences.
"Optimization and learning algorithms will have a big impact on future wearable robotic devices designed to assist a range of behaviors," says Scott Kuindersma, an assistant professor of engineering and computer science. "These results show that optimizing even very simple controllers can provide a significant, individualized benefit.”
The team next aims to apply the optimization to a more complex device that simultaneously assists multiple joints, such as a hip and ankle.
"We demonstrated a high reduction in metabolic cost by just optimizing hip extension," adds Ye Ding, a postdoctoral research fellow at SEAS. "This goes to show what you can do with a great brain and great hardware."