A commercial robotic leg could potentially benefit both higher- and lower-mobility amputees, University of Michigan roboticists have shown for the first time.
The leg provided the largest gains when the U-M team applied its own control strategy, enabling a more symmetrical gait, lower tripping risks and a reduction in strain on the study participants' sound legs and hips.
The Michigan team worked with Össur's Power Knee, provided by the company, with primary funding from the National Institutes of Health.
Multiple robotic prosthetic legs are on the market but aren't yet in wide clinical use. For many activities, the lightness and simplicity of passive legs is preferred by prosthesis users. However, for particularly taxing activities like rising from a chair, climbing stairs and hills, and walking long distances, the addition of power has the potential to help prosthesis users be more active while also preventing overuse injuries.
"The passive leg has a huge advantage in this study because the participants use it every day and are very used to its behaviors. Our study participants had just two training sessions with the powered leg. Even with that disadvantage, we observed benefits of the powered leg with both our controller and Össur's," said Robert Gregg, a professor of robotics and corresponding author of the study in the Journal of NeuroEngineering and Rehabilitation.
"Our study is significant because evidence was previously lacking for benefits of robotic knees over advanced passive knees, which is a big reason insurance companies don't typically cover robotic knees. Our results begin to provide this evidence."
In this initial study, the researchers focused on key activities of daily living in which a powered prosthesis could provide meaningful benefits compared to passive knee prostheses. Study participants repeatedly sat and stood; walked quickly on a treadmill; and repeatedly sat in a chair, walked and sat down again.
Prosthesis users who required extra walking assistance, such as a cane, found that the Power Knee offered significant help in all these tasks. Those who get around more easily on their own prostheses saw the most improvement in their gaits when Gregg's team implemented their own control algorithm on Össur's leg. One recent amputee described it as the closest they’d felt to two-legged walking on a prosthesis.
"Our goal in prosthesis control is to make the leg behave as close as possible to the missing human limb in order to prevent compensations that often lead to overuse injuries. It also matters because gait deviations can bring unwanted attention to some users," said Kevin Best, a research associate in robotics, recent U-M robotics PhD graduate and first author of the study.
The team investigated two fundamentally different control approaches. Össur's more conventional controller relies on recognizing specific features of the user's motion, indicating what they're about to do. This makes the system very safe and predictable but may not always keep up with the user's intent, Gregg said. In order to sit, users have to wait for the knee to recognize the sitting motion before it will bend, and likewise with standing.
In contrast, the control approach developed by Gregg's team continually adjusts to the user's motion. They built mathematical models of how humans move, based on large datasets of unimpaired individuals. At each moment in time, their control algorithm measures the user's thigh motion to determine the right behavior, creating more natural knee motions that are better synchronized with the user.
"With the improvements in robotic devices, it is clear that robotic prostheses offer great promise to the amputee population," said Jeff Wensman, a certified prosthetist/orthotist at Michigan Medicine and study co-author. "I am excited to see the advancement of a strategy to provide powered prosthesis users with user-synchronized control. I believe that this is a missing link to making powered prosthetics a reality for amputees."
This new control algorithm is harder to learn after using a conventional prosthesis for years, but the repeated sit-to-stand trials showed that participants were learning. Rather than getting more tired with each trial, they got faster. Then while walking, the more mobile group showed two improvements that could become significant benefits.
First, they lifted the toe of the prosthetic foot higher, reducing the risk of tripping over small obstacles or rougher terrain. Second, they didn't need to swing their hips as hard to move the prosthetic leg forward, hinting that the powered knee and lifelike control algorithm could potentially reduce back pain and allow users to go farther before getting tired, though the team couldn't measure this with the short study.
Next, the team hopes to demonstrate the safety and effectiveness of their control algorithm with stairs and ramps, followed by take-home tests. With more time to practice, the participants may be able to achieve even more. If the control strategy is successful, Össur could incorporate aspects of it into its own algorithm.
For now, the improvements with Össur's own algorithm were enough for two of the study participants to switch to the Power Knee for their everyday prosthesis, demonstrating that robotic prostheses are moving from laboratory exploration to real-world benefit.
Primary funding by the National Institutes of Health was under award No. R01HD094772, with additional support from the National Science Foundation. Össur provided initial financial support and may have a financial interest in the results.
The team has applied for patent protection for their controller with the assistance of U-M Innovation Partnerships.
Study: The clinical effects of the Össur Power Knee with phase-based and default control during sitting, standing, and walking (DOI: 10.1186/s12984-025-01729-2)
Journal
Journal of NeuroEngineering and Rehabilitation