David Quintero PhD’18 inspects a robotic leg created by the application of robot control theory. Quintero is the lead author of research conducted in the lab of Dr. Robert Gregg, Fellow, Eugene McDermott Professor at UT Dallas.

Dr. Robert Gregg has dedicated much of his life to helping stroke victims and lower-leg amputees learn to walk again. Motivated in part by a desire to help wounded veterans, the UT Dallas assistant professor of bioengineering and mechanical engineering is conducting research that is opening doors to a smoother leg motion, one that allows amputees to walk more naturally at various speeds on variable inclines.

His latest research on the subject, published online by IEEE Transactions on Robotics, describes his powered knee-ankle prosthesis that has been tested on three above-knee amputee subjects. Gregg likens this new development to horseback riding.

“Walking on a conventional robotic leg is kind of like riding a horse. You’re kind of in charge, but not really,” said Gregg, Fellow, Eugene McDermott Professor and the corresponding author of the study. “The horse does what it wants to do. You’re following the horse, but it’s the horse that’s making the steps.

“That’s not what you want when you’re walking on a leg that’s supposed to be replacing your missing limb. Our robotic leg is synchronized to the user. For the first time, the person is in charge.”

Gregg portrait

My motivation was determined in part by the conflicts in the Middle East. There was a need for robotics to be applied to amputees. Seeing how much soldiers gave to their country made me want to give back to them.

Dr. Robert Gregg, Fellow, Eugene McDermott Professor

Previous versions of these types of systems involved very sophisticated software. They were also highly sensitive to the user. No two people were programmed the same way, because each person walks a little differently. Gregg said there’s a lot of time-consuming, patient-specific customization involved in terms of the software.

“Most current methods of controlling movement are specialized, and can perform a set of predefined motions, such as walking up a small incline or up and down stairs,” said Gregg, who teaches in the Erik Jonsson School of Engineering and Computer Science. “Typically, control systems for powered prosthetic legs divide the gait cycle into several periods with distinct controllers. It requires the discrete use of a switching controller based on a set of rules and dozens of control parameters that must be tuned across users and activities. The controller has to interpret what the user is doing.”

To improve the amputee’s ability to walk naturally, Gregg synchronized the progression of the prosthetic to the motion of the user in real time instead of choosing from a set of options. This unified the gait cycle and provided a fluid stride that allowed three above-knee amputees to walk naturally at several speed and slope conditions.

In this new approach, the amputee would be using his or her hips to coordinate the control of the knee and the ankle, injecting and removing energy into the gait cycle at different times. This produces the desired movement at the joint.

“The ‘secret sauce’ is the sensors,” Gregg said. “The sensors measure the angle of the hip. When I’m walking, my hips are moving back and forth, rhythmically. The hips act like a crank that moves the leg forward. The sensor measures where that crank is at any point and time. The simplicity lies in having the knee and the ankle follow the crank of the hip.

“By using a new way of controlling the leg, you don’t have such a complicated mess of parameters in the software. This is essentially plug and play from one person to the next. Subjects can also walk faster and on inclines with this leg.”

Gregg said his innovation came after studying the human body.

“We did studies with non-amputees and found that when they were walking, the hip motion was highly correlated to the timing of the knee and the ankle motion. The hip predicts the knee and the ankle motion,” he said.

A non-amputee tests a powered knee-ankle prosthesis. If you don't see the video, watch it on YouTube.

Gregg recently presented the next iteration of his design at the IEEE International Conference on Robotics & Automation in Brisbane, Australia. Dubbed the Gen2 UT Dallas Powered Knee-Ankle Prosthesis, the device has high-torque motors and low-ratio transmissions. It also has a custom gear that provides for free-swinging knee motion. It is quieter and can regenerate energy like an electric car. Gregg has begun testing the new device and expects amputee trials to begin this summer.

Gregg said his long-term goal is to create a powered prosthetic leg that can do almost anything the human leg can do without being limited by a set of preprogrammed responses. He said his work has piqued the interest of a robotics company that shares his vision.

“My motivation was determined in part by the conflicts in the Middle East,” he said. “There was a need for robotics to be applied to amputees. Seeing how much soldiers gave to their country made me want to give back to them.”

Other researchers who contributed to the journal paper are lead author David Quintero PhD’18, Dario Villarreal PhD’17, electrical engineering graduate student Daniel Lambert, and Dr. Susan Kapp at the University of Washington. Research assistant Toby Elery, research scientist Dr. Siavash Rezazadeh, computer engineering student Jack Doan, Eugene McDermott graduate fellow Hanqi Zhu MSEE’17 and research engineer Christopher Nesler contributed to the Gen2 UT Dallas Powered Knee-Ankle Prosthesis.

The research is funded by the National Institutes of Health, the National Science Foundation, the Burroughs Wellcome Fund, and the Mexican National Council of Science and Technology.