Robo-Legs Provide Insight into Human Gait
October 2012 Issue
Developing O&P devices that restore 100 percent of lost human function and mobility following an amputation or debilitating injury has been a long-time goal in O&P product development, and studying gait patterns is an essential, foundational component in achieving that goal.
Robotic engineers at the University of Arizona (UA), Tucson, have taken steps toward advancing that goal by building a pair of legs that mimic the mechanical and neurological systems that allow humans to walk upright. The result is a robot with an almost jaunty, rhythmic gait. What researchers learn from these robo-legs could lead to a greater understanding of exactly how humans walk and eventually to recreating the complex process for patients who cannot walk, perhaps through the development of advanced O&P devices.
One human step requires 30 major muscles in each leg to contract or relax at just the right time in just the right order. The whole sequence—and the millions of tiny adjustments needed to maintain balance and navigate changing terrain—is coordinated by a neural network in the lumbar region of the spinal cord. This central pattern generator (CPG) is what makes the sophisticated task of walking a more-or-less unconscious activity for most able-bodied adults and something babies naturally learn with practice.
In contrast, most legged robots use motors to run their joints individually. These motors must be powered all the time, even when the non-mechanical joint would naturally rotate passively, and require a central processor to coordinate their movements. The result is a gait that is, by definition, mechanical.
A research team led by M. Anthony Lewis, PhD, and Theresa J. Klein, PhD, took a different approach—one that is much more biologically accurate. They connected the hip, knee, and ankle joints of their robotic leg with Kevlar straps that function as the muscles that alternately flex and extend the joints—or allow them to swing freely.
The "neurorobotic" model includes biarticulate muscles that span two joints—the gastrocnemius, rectus femoris, vatus lateralis, and hamstring muscle group—and a series of nine actuators that cause them to contract or relax.
The actuators are controlled by a simple half-center CPG. As the CPG fires two signals alternately, it generates a rhythmic stance/swing step cycle. Load sensors in the legs mimic the Golgi tendons in human legs and allow the CPG to adjust the firing rhythm for stability across a range of walking speeds. "We close the feedback loop by allowing the mechanics of the body to generate sensory signals that entrain the CPG and generate motion through reflexes," Lewis and Klein wrote in the July 5, 2012, issue of the Journal of Neuroengineering.
In addition to producing a more human-like gait for robots, the legs are more energy-efficient than motorized versions.
"Walking actually uses very little energy," Lewis says. "Biarticulate muscles have an almost mystical power to transfer power from the hip to the ankle, so the energy load is distributed over the entire leg. Motorized joints can't match the efficiency of the agonist/antagonist action. It's difficult to recreate in a robot, but it's intriguing."
These UA legs are the first phase of a complete humanoid robot called Achilles. The next step is to add a torso on top of the legs, which will then have to learn to balance while walking with an even more natural gait. Simple enough for a human baby, but a huge challenge for a robot.
"We have our work cut out for us," says Lewis, who has a doctoral degree in electrical engineering and neuroscience from the University of Southern California, Los Angeles, and an undergraduate degree in cybernetics from the University of California, Los Angeles.
Lewis says he became interested in using his skills as a roboticist to explore biological ideas early in his doctoral program. He credits work at Case Western Reserve University, Cleveland, Ohio, during the 1990s that combined biology and engineering for a lot of his inspiration.
Other researchers have also explored creating artificial walking muscles. Ryuma Niiyama, PhD, of the University of Tokyo, Japan, who is now with the Robot Locomotion Group at the Massachusetts Institute of Technology, Cambridge, Computer Science and Artificial Intelligence Laboratory Center for Robotics, presented a bipedal robot perched on Cheetah running blades in 2010. The robot, called Athlete, has seven sets of pneumatic-powered muscles above the knee in each leg and touch sensors on each foot. It ran about a yard before it lost its balance. Niiyama said his purpose was to better understand how humans control their muscles while in motion. The ultimate goal, he said, is to create more efficient prostheses.
The Achilles project has been under way for about ten years, according to Lewis.
"I started with a small robot to test the idea of the CPG," he recalls. "Then we'd add something else to test and build a new robot."
The hard part is that it takes time to build robots by hand and to find students with the right background who want to invest the time in such a long-term project. Lewis, who was Klein's advisor as she built the Achilles legs over four years, says it also takes a long time to get student researchers up to speed. "Whenever you're dealing with anything physical, things break," he says, "so you're always fixing things. About 90 percent of our research is about dealing with problems that are not of earth-shattering importance."
While the team worked on Achilles at UA, the university provided only partial funding for the project. The rest of the support came from individual grants that the researchers secured themselves.
Lewis' previous work has also been funded by the Office of Naval Research, which he says is very interested in the CPG portion of the research. "They are particularly concerned with how it might help veterans returning from war with spinal-cord injuries (SCIs)," he explains.
But otherwise, Lewis says the funding climate isn't very strong for theoretical research. "We actually completed the legs on our own dime."
As a result, Klein has gone to work for an aerospace company after completing her doctoral degree. Lewis also recently left the university to head up his own company, Iguana Robotics, Tucson. He has about six employees and will initially be competing for grants with his former university colleagues to complete the Achilles project, but says that he plans to seek private investors in time.
Right now, the purpose of Lewis' work is to expand the base of knowledge in the fields of robotics, neurophysiology, and biomechanics, rather than to produce an end product. Eventually he would like to build things that are practical and useful, such as robots to send into dangerous situations in place of people, or products to help patients with SCIs recover the ability to walk. Lewis says he also envisioned someday adapting the robots as simulators for therapists training to work with SCIs or using them as ambulatory assistive devices. They could also become the basis for the ultimate powered prostheses.
"It would be great if we could create a lower-limb research program like DARPA [Defense Advanced Research Projects Agency] did for upper limbs," he says. But such uses are many, many baby steps down the road.
Kate Hawthorne is a freelance writer living and working in Fort Collins, Colorado. She can be reached at
Editor's note: To read the research paper on the Achilles project, visit www.oandp.com/link/173