at this point the availability of robotic prosthetic limbs comes down to one thing, how much money/skill do you have. servos, sensors, microcontrollers, cnc fabrication machines, pcb mills, all off the shelf parts. seen some pretty cool artificial appendages on many a maker site. only thing we still dont have is a neural interface, or do we? medical researchers are always making advancements with brain-implantable electrode array chips that have allowed the blind to see and quadriplegics to do things other than looking around.
One of my buddies is a ChemE looking at med school, and did Neuroscience for a couple years. There is a lot of research into bionic limbs with a right and proper neural interface, but he tells me that there are some problems. Apparently it's not as simple as using electrodes to connect one's nervous impulses to the prosthesis' wires; I don't recall all the details, but there are apparently synchronization and timing issues with systems that work with a direct connection.
timing issues yes but thats not something that cant be delt with in software/hardware. output simply requires the electrode grid to pick up the impulses over an area of the correct part of the brain. each electrode reads a large cluster of nerve cells, not individual neurons. were not yet at a point where we could connect one wire to one neuron (you would need some pretty advanced nanotech for that). the resolution and size of the grid is definitely a factor. the more electrodes and the more area is covered, the more information the computer has about what the operator is trying to do. its sorta like a ccd in a digital camera. of course more electrodes means more electronics, since you need an adc to read each electrode (neural impulses are mostly analog). also each electrode will likely need some analog processing circuitry in addition to an adc, and some memory to buffer the data while it waits for the cpu to process it. you could put all this stuff on one chip, a simple feat of semiconductor fabrication (neural interface chip is not an off the shelf part, yet). you also have packaging, the chip itself would need support parts, power (induction power coil, battery, glucose fuel cell that runs on blood) and a means to communicate (rf transceiver, through the skin optical interface, trans-dermal wiring) with an external receiver, all in a bio-resistant package small enough to fit in yer skull.
the heavy work is actually done on the cpu. it needs to associate patterns of impulses with actions. these patterns are likely subtle and contain information about what exactly you are trying to do with your missing limb. the real work is training both the computer and the patient to understand each-other. the computer needs to build a database of impulse signatures and mappings to actions to be carried out by the actuators, while the patient needs to train their brain to be consistent in its impulses, mainly through trial and error. you also have issues with processing speed, you will have some latency between the impulse and the computer figuring out what to do with it.
ive seen several documentaries on the subject, and what ive seen seems to indicate that the technology is viable for the most part. there are still bugs to hammer out of the technology, and its still fairly crude (like seeing human guinea pigs with a large metal apparatus potruding from their skulls) but what has been done so far is pretty cool.
