█ JULI BERWALD
A brain-machine interface is the linkage of the brain to a mechanical device exterior to the body in such a manner that the device is controlled by natural signals from the brain. An important goal for developing such technology is to aid people who are paralyzed or otherwise physically impared. The military has interest in brain-machine interfaces as a means of controlling robotics from a distance with extreme accuracy and precision.
One of the major technological hurdles in the development of brain-machine interfaces is the understanding of neural patterns required to accomplish tasks. One company headed by American scientist Phillip Kennedy has made great advances in this area. Kennedy has developed a very small neurotropic device that is implanted into the motor cortex of the brain of severely paralyzed people. This device transmits electronic signals from the person's brain to electronic equipment that then translates the signals to a computer. People with the implant learn to control a mouse on the computer and to type text using electronic signals in their brain.
The extension of this technology is the understanding of the neural patterns required to control complex motor tasks. In 2000, scientists at Duke University implanted an array of 96 electrodes into the brain of an owl monkey. The electrical signals measured on each of the electrodes were collected when the monkey performed certain tasks, including reaching for food. These signals were then analyzed and mathematical algorithms were developed that allowed scientists to predict the trajectory of the monkey's hand from the neural signals. The scientists then programmed a robotic arm to move in three dimensions according to the monkey's brain signals. They eventually transmitted these signals over the Internet to a laboratory at MIT, where another robotic arm 600 miles away was controlled by the monkey's neural signals.
The Defense Advanced Research Projects Agency (DARPA) is extremely interested in brain-machine interfaces for controlling robotics and interpreting sensory information. In 2001, they authorized funding for the Brain-Machine Interfaces program. The goals of this program are to create new technologies that enhance human performance through non-invasive integration of neural signals into external devices. This includes understanding the neural codes required to complete complex motor tasks, building a feedback loop from an external device back to the brain, and fabricating new materials required to capture neural commands. In addition, biomimetic systems that integrate nueral signals are of interest.
Other defense related projects investigate neural networks and optics. Scientists at the U.S. Army Aviation and Missile Command (Weapons Sciences Directorate) headquartered at the Redstone Arsenal, Alabama are working intently on projects designed to integrate optic "flow" and automatic target recognition systems. These projects utilize mathematical techniques improving image factorization (e.g., image decomposition). For example, neural network based optics using specific algorithms can translate optic flow into four separate image planes that represent various motion parameters. In addition to targeting, neural network based optics may be used to navigate autonomous vehicles and other robotics.
█ FURTHER READING:
Defense Advanced Research Projects Agency: Defense Sciences Office, "Brain Machine Interfaces" <http://www.darpa.mil/dso/thrust/biosci/brainmi.htm> (March 26,2003).
Neural Signals <http://www.neuralsignals.com> (March 26, 2003).
Science Daily: "Monkeys Control A Robot Arm Via Brain Signals" <http://www.sciencedaily.com/releases/2000/11/001116080512.htm> (November 16, 2000).
Biological and Biomimetic Systems
DARPA (Defense Advanced Research Projects Agency)