Engineers with the University of Arizona have developed a set of robotic legs that mimic the movements of humans, according to a report from Fox News. Theresa Klein and M. Anthony Lewis, both of the University’s Department of Electrical and Computer Engineering, have described their project in a new paper. The pair’s robot models the “neuromuscular architecture of human walking,” according to their article in the Journal of Neural Engineering.
Their robot has been remarkably successful at achieving simple walking on a treadmill, a feat accomplished by incorporating a new model of neural networking in the limbs. The pair’s robot incorporates a “half-centre” neural network into the robots hips, which operates on simple signals from the limbs to produce the alternating motion of the walking legs. This network, which replicates the nervous network present in human legs, has never before been incorporated into a walking robot.
“We’ve tried various robot systems. But if we build special-purpose robots, when we modify trial procedures or switch to different projects, those robots become useless. Also, developing robots is very time-consuming. So, we wanted to develop a robot that can do what people do, using the same tools people use. That’s why we’ve developed Mahoro,” say scientists.
Ms. Klein, a Ph.D. student at the university, says, “Interestingly, we were able to produce a walking gait, without balance, which mimicked human walking with only a simple half-centre controlling the hips and a set of reflex responses controlling the lower limb.” The team notes that this model could help explain how human babies begin to walk even without a fully-developed sense of balance and coordination.
The half-center system works by detecting the force of each step and then sending that reflex signal to a junction which initiates the next step. The movement itself is achieved by means of motors in the thigh and calf counterparts of the robotic leg, which exert force on Kevlar straps that replicate the function of leg muscles. Electronic sensors attached to the straps model the neural sensors in human legs that initiate changes in motion by detecting the force applied to the stretched and contracting muscle.
Each leg of the university’s robot consists of a hip, knee and ankle moved by nine muscle actuators. Muscle contraction is mimicked by rotating the motor to pull on Kevlar straps. Each muscle strap features a load sensor that models a tendon in a human leg, sensing muscle tension when a muscle is contracted and sending signals to the brain about how much force is being exerted and where.
In their paper, the team notes that their model seems to mimic the natural walking motions of infants, but that this motion appears to be replaced or augmented by “learned” walking behaviors as individual’s gain walking experience as they age. Their research has applications both for understanding the development of muscle-sensory interaction in human development, and in the construction of better walking robots: “This robot represents a complete physical, or ‘neurorobotic’, model of the system, demonstrating the usefulness of this type of robotics research for investigating the neurophysiological processes underlying walking in humans and animals.”
The research appears in the July 6 issue of the Journal of Neural Engineering.
