The VICE Channels

    This Nerve-Controlled Robot Leg May Be the World's First

    Written by

    Derek Mead


    As advanced as prosthetics now are, they still can't beat the real thing. A team of engineers has, however, taken a big step forward by wiring a robotic leg directly into a human nervous system for perhaps the first time.

    The video above shows the 32-year-old patient, who lost his leg in a 2009 motorcycle accident, walking up and down a ramp and some stairs, which highlights just how smoothly the prosthetic works. While the prosthetic itself is clearly quite advanced, its control system is the star of the show.

    According to work published in the New England Journal of Medicine, an engineering team at the Rehabilitation Institute of Chicago decoded electromyographic (EMG) signals, which are produced by nerves in skeletal muscles as your brain controls the way you move, using a pattern-recognition algorithm.

    After redirecting some of the man's lower-leg nerves to instead fire in his thigh, the team picked up those muscle-control signals with sensors built into the prosthetic to control its movement. In other words, the man's brain signals directly control the robotic leg's movement. Biomedical engineer and team lead Levi Hargrove told Nature that, as far as he knows, it's the first time such interfacing has been deployed successfully in a device that controls both knee and ankle articulation.

    Currently, even the smartest prosthetics rely on muscle movements or other forms of outside control. A few months ago, Motherboard met Nigel Ackland, whose bebionic3 robot hand is one of the most advanced prosthetics in the world. He was more than kind enough to show it off for us, and its ability to rotate at the wrist and grip in various configurations, including closing and extending individual digits, was truly impressive.

    Ackland's arm is controlled by muscles in his forearm; flexing one way activates a sensor that causes the prosthetic to respond. Ackland was impressively adept at controlling his prosthesis, but that verb—control—betrays an inherent aspect of even highly advanced prostheses: patients still need to learn how to use them adeptly. And no matter how mellow the learning curve is, it's still there.

    Integrating a prosthetic directly into a patient's nervous system would seem to be the answer. Rather than learning how to use one's robo hand, it could be controlled just as directly as your flesh-and-bone models. The concept has been proven before; research earlier this year showed that a paralyzed person could control robot limbs with her thoughts, but actually deploying it in a prosthetic—especially one as challenging as a lower leg—is even harder. So far, the proof of concept appears to be working (walking?) well, and Hargrove's team hopes to have it ready for broader use within a few years.