Weekly Robot Update 008

No wheels or legs on this one. It doesn't even have it's own brain. And while it doesn't strictly qualify as a "robot", the main stream Internet press continues to refer to it as such. I figure that's good enough for me. It is the proof of concept for a Brain Machine Interface or BMI. It's a mechanical hand that responds directly to human thought, and it is this weeks robot.

The Brain Machine Interface was developed jointly by the Honda Reaserch Insititue Japan, and the Advanced Telecommunications Research Institute International. The machine is an analog of a typical human hand. Five digits, an opposable thumb, and articulated joints in the right places give it comparable functionality to your own hand. The hand is statically mounted to a stand which is connected to specially designed computer. The computer is connected to a medical grade fMRI. The hand's purpose is simple; Mimic the motions of its human controller in a game of rock/paper/scissors. What's unique about this process is that there are no controls to speak of. The operator simply makes a particular gesture, say the sign for victory, and the fMRI reads his brain activity. This is then sent to the hand which precisely duplicates the same hand sign.

Other researchers around the world are working on similar concepts for machines controlled by thought alone. But those projects often involved risky surgical procedures directly on the brain of the test subject. Using the fMRI, Honda and ATR scientists are able to get the same results without drilling holes in anyone's head. In addition, most Brain Computer Interface prototypes require significant training to use even after the device has been implanted. But with the new BMI, operators can start playing robotic rock/paper/scissors within minutes. No risky implants, no extensive training, and nearly perfect responses make even this fetal stage of BMI a cut above the rest.

There are however some drawbacks to this approach. The equipment needed to record the brains activity is unwieldy to say the least. it takes up the better part of a room and requires the subject to lay inside of it. To make this at all a usable technology, the fMRI's size and weight will have to be decreased by several orders of magnitude. In addition, improvements in the devices scanning resolution will have to be improved to communicate more complex instructions to a robot or prosthetic. There is also the matter of the seven second lag between the controller's thought and the machines response. This is a much simpler problem to solve. it simply requires a better and faster computer between the fMRI and the robot. Researchers believe that they will eventually be able to wire up a connection that can be processed so fast that it will outpace human reflexes. Think of moving your hand, and the robot will actually be quicker on the draw than you are. Have a look at the video for a real world demonstration of the BMI in action.

The BMI project holds out strong hope for those that have lost or been born without limbs. A prosthetic arm wired up through BMI could eventually function even better than a human original. And once the technology is developed further, there is no reason to stop at prosthetics. Why not drive your car, or fly a plane, or operate a computer using only your mind. Project scientists estimate that it will be five years or so before we start seeing useful technology built on BMI. But for something as promising as thought controlled machines, a half decade is scarcely a wait at all. [via New Scientist Tech]


“A man paints with his brains and not with his hands.” - Michelangelo