Scientists have taken the first tentative steps toward developing a real
”bionic man,” one whose nerve signals would be analyzed and processed microelectronically so that a person`s natural nerve impulses could control a robotic hand as if it were his own.
The research, although still at least a decade away from application in humans, is financed by the U.S. Department of Veterans Affairs to promote technology that may restore normal function to people who have lost hands in combat or accidents.
The technology could one day go beyond helping the handicapped to giving superpowers to ordinary humans, researchers predict.
Bernard Widrow, a Stanford University electrical engineering professor, and his colleagues have demonstrated that a ”nerve chip,” incorporating living nerve tissue into a silicon microchip, can be implanted and will work for a long period within an animal.
As envisioned by researchers, a person with a missing hand will one day be fitted with several such chips, implanted within the stump of a damaged limb to encircle the nerves that once led to the severed hand.
An external computer would send to the implanted chips and receive from them electrical impulses, writing an individual program to help the patient control his personalized robotic hand. This would be accomplished by taking advantage of the well-documented ”phantom limb” phenomenon, in which a person who has lost a limb, hand or foot still senses the presence of the missing part, even to the point of feeling heat, cold or pain.
An image of a hand on a computer screen would instruct the patient to open his hand, make a fist, move his thumb and so on. The patient would follow these instructions, watching the hand on the screen move and sending nerve impulses to control his ”phantom” hand, just as he did before it was lost.
The computer would chart individual patterns of neural impulses and construct the program that would later be downloaded into control electronics embedded in the robotic hand for that individual.
A similar interactive learning process would be used to determine what electronic signal patterns would provide sensory tactile feelings to the individual from his robotic hand.
To date, scientists have implanted microchips containing numerous holes into the feet of rats where nerves had been severed.
Individual nerve fibers, called axons, regenerated through the holes like string beans growing through chicken wire, said Greg Kovacs, a Stanford graduate student in medicine and electrical engineering.
Iridium microelectrodes in the chip monitored electrical signals generated by the rat`s nerve system and sent through individual axons. The microelectrodes also stimulated axons with outside electrical impulses. This demonstrated these chips` potential as a human-machine interface, tying motor and sensory nerve impulses to an electronic computerized control system.
Stanford`s accomplishment is a tribute to materials engineering and microtechnology. The chips are made with holes about 8 microns wide so that axons that are 2 to 6 microns can grow through them. A micron is one-thousandth of a millimeter. A human hair is about 100 microns across.
Each chip was coated with silicon nitride to protect the microcircuits from degradation caused by body fluids.
The next step is to design a microchip that can interact with hundreds or thousands of axons and to determine the best design to get maximum regrowth of nerve axons through holes in a single chip, Kovacs said.
Stanford`s integrated circuit lab uses commercial plasma etching technology to make the hole-laden chips, avoiding exotic, ultra-expensive techniques. Stanford will share its chips with researchers at other centers across the country and even make chips to their specifications to stimulate bionic research, Kovacs said.
A major task during the next decade will be mapping neural electrical activity, first in animals and eventually in humans. Scientists must determine how to collect and average enough signals to generate synthetic impulses that can control robotic hands.
They also must learn to translate electronic signals from sensors implanted in robotic hands to stimulate the human nervous system with feedback information, telling the brain when the hand`s index finger is touching the thumb and ”feeling” a cup of hot coffee when the hand grasps it.
”You have to find some way of mapping from thousands of circuits down to just a few circuits,” Stanford`s Widrow said. ”And there`s no road map there; there`s nothing that tells you which of the living circuits previously served which function on the hand once the hand is gone.”
And computer neural systems that learn from experience must be developed, he said. Several mathematical strategies for accomplishing this are in the works.
The hope is that a computer-smart electronic neural network can detect averages and patterns among many organic neural axons doing related jobs. Scientists expect they can run a human-machine communications system without understanding all the subtleties and nuances of the nervous system.
”You shouldn`t need to know how the whole nervous system works to communicate with it,” Kovacs said.
Within three years, the Stanford researchers believe they will discover whether this approach is feasible.
Even before bionic technology is used to give amputees natural control over robotic hands, it may find a role in helping patients whose hands are severed and then surgically reattached.
When surgeons sew major nerves back together, the individual axons regenerate in a scrambled fashion, often reuniting with fibers that had been connected to different axons before they were severed. This axon scrambling makes it difficult for a patient to use his resewn hand, a problem only partially overcome by rehabilitation therapy.
Implanting microchips for the axons to grow through would provide a potential way to descramble this jumble by rerouting nerve signals to the appropriate pathways, Widrow said.
Although scientists have planned a research path they expect could lead to bionic help for amputees in the next century, they caution against raising hopes prematurely.
”There are still many technological obstacles to overcome before we even think about doing this work in humans,” Kovacs said. ”Giving robotic hands and legs to amputees and paraplegics is a project that could cost as much as putting a man on the moon.”
Long-range applications could go beyond helping the handicapped to giving ordinary humans superpowers, the scientists say.
Widrow listed some of these possibilities while describing his team`s research in a plenary lecture to the International Joint Conference on Neural Networks earlier this year.
The human-machine interface pioneered by Stanford researchers might one day allow jet pilots to plug their nerve impulses directly into their airplane`s control system, eliminating the need to touch controls with their hands.
And the technology might give a person a way to directly feel sensations stimulated by phenomena such as radiation that now are totally undetectable to human senses.
