New Control Algorithm Enhances Robotic Knee Prosthetics
A collaborative effort between researchers at the University of Michigan and medical device maker Össur is delivering a breakthrough in powered prosthetic knees. The team tested a novel control algorithm on Össur’s Power Knee, funded in part by the National Institutes of Health. Early results show meaningful gains in gait symmetry, balance, and the distribution of load between the sound leg and hip, across participants with varying mobility levels.
While multiple robotic prosthetic legs are available commercially, their adoption in routine clinical use remains limited. Many users prefer passive, lighter prostheses for daily activities. Yet for taxing tasks—rising from a chair, climbing stairs or hills, and walking longer distances—a powered knee offers the potential to expand activity levels while reducing overuse injuries.
Why the New Controller Matters
“The passive leg has a huge advantage because participants rely on it daily and are familiar with its behavior,” said Robert Gregg, a robotics professor at the University of Michigan and the study’s corresponding author. “Even with just two training sessions on the powered leg, we observed benefits with both our controller and Össur’s.”
Gregg’s team illustrates a key point: evidence supporting the superiority of robotic knees over advanced passive knees has been scarce. By providing measurable improvements in tasks of daily living, the study helps address a major hurdle for insurance coverage and patient acceptance.
Two Control Strategies, One Goal
The researchers compared two approaches to steering the powered knee. Össur’s conventional controller relies on recognizing distinct motion features to predict user intent. This makes the system safe and predictable but can introduce lag, especially when the user transitions between sitting and standing. In practical terms, users may wait for the knee to recognize their intent before bending, which can feel stilted.
In contrast, Gregg’s group developed a continuously adapting controller. Drawing on mathematical models of human motion and large datasets from able-bodied individuals, the algorithm gauges the user’s thigh motion in real time to determine the knee’s behavior. The result is more natural, synchronized knee movement and reduced compensations that often lead to overuse injuries.
Early Findings and Real-World Impact
During the study, participants repeatedly performed sit-stand cycles and walked on a treadmill, followed by cycles that included more sit and stand actions. Those who relied on additional walking support, such as a cane, benefited notably from the Power Knee. Even among participants already comfortable with passive prostheses, the powered knee—guided by Gregg’s algorithm—produced smoother gaits and improved toe clearance during walking.
One participant described the powered knee as “the closest you can feel to two-legged walking on a prosthesis,” highlighting the human-like motion that the controller promotes. The team emphasizes that the goal is to emulate the natural dynamics of a missing limb to prevent compensatory patterns that jeopardize long-term joint health.
Looking Ahead: Safety, Training, and Clinical Use
Next steps include trials focusing on stairs and ramps, followed by take-home testing to gauge real-world durability and user satisfaction. If the controller proves safe and effective, Össur may integrate elements of the new approach into its commercial algorithms, accelerating broader adoption.
“The improvements we observed with our controller suggest powered prostheses can become a practical reality for more amputees,” said Kevin Best, a robotics researcher and coauthor. “This is a missing link toward making powered knee prostheses a routine option.”
Broader Implications for Prosthetics
The study ultimately demonstrates that advancing control strategies is as crucial as hardware development. By aligning device behavior with human motor intent, powered prostheses can reduce fatigue, lower injury risk, and support higher activity levels for a broader spectrum of users. This progress could help shift payers’ and clinicians’ views toward more active adoption of robotic knees in everyday life.
Primary funding came from the National Institutes of Health (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 research team has filed for patent protection for their controller with assistance from U-M Innovation Partnerships.
