A Better Way To Make Mind-Controlled Prosthetic Limbs 28
the_newsbeagle writes: To make a brain-machine interface, you need a way to capture neurons' electric signals. The most precise and most invasive way uses implants that are stuck in the gray matter. The least precise and least invasive way uses EEG sensors stuck to the scalp. But researchers at Johns Hopkins University say there's a third way that gets the best of both worlds, which is not too invasive and fairly precise. They use ECoG systems, in which a mesh of electrodes is placed under the skull, draped over the surface of the cortex.
They're testing their systems on epilepsy patients, who have these ECoG systems inserted anyway while they're waiting for surgery (the electrodes record the source of their seizures). The researchers are capturing these patients' movement commands from their brains, and using them to control robotic limbs. Someday such a system could be used by amputees to control their prosthetic limbs.
They're testing their systems on epilepsy patients, who have these ECoG systems inserted anyway while they're waiting for surgery (the electrodes record the source of their seizures). The researchers are capturing these patients' movement commands from their brains, and using them to control robotic limbs. Someday such a system could be used by amputees to control their prosthetic limbs.
Re:Not too invasive? (Score:5, Informative)
Re:Not too invasive? (Score:2, Informative)
Not that tiny, I did neuroelectrical design for over a decade. Electrodes outside the local motor nerves themselves, have a *ridicious* amount of noise from irrelevant neural material. Even electrodes *on* or in the nerve have ridiculous amounts of noise. And worse, as soon as some fumbfutz tries to digitize and sample it, they throw out most of the data. The signals are not like binary signals, on or off. They're pulse driven. state changes in the right amount, on the right nerves, trigger motion or sensation. The electrical equivalent my monitoring the brain is like monitoring an intersection by listening to the traffic. You've already spent a lot of time gathering information, and by the time you have enough data to be certain of the signal, it's too late and too gross a measurement for detailed mechanical manipulation.
Oh, and the smaller the electrode to help isolate the signals, the more electrical noise you pick up from being stuck in a saline medium, trying to read electrochemical signals in a conductive medium. It's an unavoidable problem until and unless Dr. Rosen's ar some other work succeeds in create an electrode array that transects the nerve and has the nerve cells actually feeding *through* the array.
I'm afraid similar work was done with myo-electric signals to control the original "Boston Arm", which used skin electrodes on the shoulder of the missing arm. The phase delay necessary to sample and collect a valid signal is too large: that's one of the reasons the Boston Arm never became popular.
Mind you, it was cool as futz. I talked to the engineer who did the mechanical design, who actually worked for an insurance company. When I asked him how much it could lift, he said "12 ounces". When I asked him why, he brought his beer bottle to his mouth....