NEW YORK, June 7 (UPI) -- Researchers in New York City are teaming with the U.S. Navy and scientists in Russia to build electronic circuits that mimic the brain, producing an agile controller that can maneuver robot vehicles with speed and precision.
The devices are based on a circuit in the cerebellum, the part of the brain that helps organize the body's motions. Specifically, the new technology imitates the olivocerebellar circuit, which controls balance and limb movement.
"It's present in all vertebrates -- it's very much the same from the most simple to the most complex brains," researcher Rodolfo Llinas, a neuroscientist at New York University Medical School, told United Press International.
"The assumption is that it is conserved (in evolution) because it embodies a very intelligent solution," Llinas said. "As the system is involved in motor coordination -- and we want to have a machine that has sophisticated motor control -- then the choice (of the circuit to mimic) was easy."
For the past three years, Llinas has partnered with colleagues at Nizhny Novgorod State University in Russia to implement an electronic version of the olivocerebellar circuit.
"Controls in robotics are for the most part algorithmic," he explained. "It's basically software, and the software instructions are written in a particular order -- you follow a particular set of steps."
In addition, the computations are contained in a system that is distinct from the one it controls.
"The nervous system, on the other hand, is not algorithmic," Llinas said. The same cells that gather the sensory data from the muscles also have a key role in operating the muscles as well, so both sensory and motor systems are wedded together, "unlike what happens in digital computers."
So the researchers are developing analog circuits "that can control complicated systems faster, smaller and with less complexity than digital systems. If you tried a digital system, you find very quickly the system can bog down, because (it) can only scale (up) so far without becoming ungainly," Llinas said.
The new controller, like the olivocerebellar circuit, is made up of clusters that interact electronically with one another. The interactions are in the form of oscillations -- electronic waves with regular peaks and valleys. When the controller issues a command for a movement, instead of changing the frequency or magnitude of the peaks and valleys, the command merely shifts the phase -- the position of the peaks and valleys -- which requires less hardware for the same result, Llinas explained.
"Instead of a whole lot of computations by a whole lot of elements, a few elements can work alone very, very fast ... If the models turn out the way we think they will, they should be able to control very complicated systems, yet be very simple," he said.
"This technology could impact land, air and underwater autonomous vehicles. There are both military and civilian applications," Thomas McKenna, program officer with the Office of Naval Research, told UPI.
Future robots could benefit from the highly precise "brain control" of movements this novel system could grant, benefiting "surveillance, rescue, and other operations while reducing the risks to humans of hazardous conditions," said McKenna, who also is a computational neuroscientist.
Such systems also could help disabled people, said Ferdinando Mussa-Ivaldi, a neuroscientist at Northwestern University in Chicago.
"Think of a paralyzed patient. It is possible to imagine that many ordinary tasks -- such as getting a glass of water, dressing, undressing, transferring to a wheel chair -- could be carried out by robotic assistants, thus providing the patient with more independence. Such machines would need to learn a huge repertoire of actions and adapt these actions both to the patient and to the environment," Mussa-Ivaldi told UPI.
"Currently available technologies are not adequate for this kind of scenario," he added. "The research of Llinas and coworkers can lead toward this goal by developing computational hardware that imitates the brain."
This summer, the researchers plan to test the controller in an undersea, mobile autonomous research vehicle, or MARV, at the Naval Undersea Warfare Center in Newport, R.I. They will attempt to maneuver the MARV in and out of a docking tube by delicately manipulating its actuators, or "muscles," to change the direction and speed of the robot.
"The immediate military application being explored is quiet and precise maneuvering of underwater vehicles for surveillance," McKenna said, adding the team also is exploring the system for underwater operations close to shore.
"These results will influence whether we then focus on developing a better prototype underwater vehicle or look at other means of implementing this controller or a modified controller," McKenna said. "If we pass the initial technical hurdles, it could become operational in the near term."
The controller could help "anything that moves, anything that needs control," Llinas said. "If one can have devices that are small and reliable that can control many variables in a coordinated fashion, this would be an industrially attractive solution to digital control when speed and expense are an issue."
Mussa-Ivaldi said "the most important step now appears to be that of confronting ideas and prototypes with 'real life' tasks, such as controlling the navigation of a vehicle in a poorly known environment, or bringing food to the mouth of a patient."
Many challenges remain, however, he added. "Some (challenges) concern the ability to create a repertoire of elementary actions from experience."
For example, Mussa-Ivaldi said, an engineer may attempt to control a complex aerial maneuver of an aircraft by employing a complex mathematical model of both the air and the aircraft.
"A bird does the same by learning from little mistakes and successes," he said. "The machines of the future will have to do like the birds."
--
Charles Choi covers research for UPI Science News. E-mail [email protected]
