Monday, March 8, 2010

Prosthetics: moving right along


The Defense Advanced Research Projects Agency (DARPA) mounted a major effort to improve prosthetic limbs back in 2000. According to Wired magazine, they've made significant progress.

DARPA, the military’s risk-taking research agency, is launching the next phase of its Revolutionizing Prosthetics program, which was started in 2000 with the goal of creating a fully-functioning, neurally-controlled human limb within five years.

Since then, the agency has made plenty of progress. They’re currently doing human trials of the DEKA Arm [shown below], a prosthetic that allows users to complete day-to-day tasks with unprecedented ease.




That arm uses a joystick-style interface, with a user tapping commands with their toes to trigger movements with the arm. At Johns Hopkins, DARPA-funded researchers are still working on an arm that uses a 100-sensor neural interface to create a brain-body meld much like what’s inherent in natural limbs.

But although DARPA had hoped to have a fully-functional, neuro-prosthetic model ready by 2010, the agency’s researchers have yet to master the integration of human neural pathways with artificial platforms. For one, neural-recording interfaces are notoriously short-lived, with a life-span of around two years, and they don’t extract adequate information to yield seamless movement from brain to neurons to limbs. A seemingly simple motion, like using an arm to eat, is actually a series of thousands of movements, sensations, cues and brain-neuron communications. Right now, DARPA’s prototypes can transmit 500 events per second. According to the agency, that’s not nearly enough.

So DARPA’s launching a new program, Histology for Interface Stability Over Time, in hopes of creating not only a neurally-controlled limb, but one that has a 70-year lifespan and flawless integration with the human body.

It’s a three-year, three-phase initiative that’s first and foremost about failure. DARPA wants to know why neural-recording interfaces are apt to break down or suffer lagging performance, and how researchers can predict that failure sooner — before an amputee is stuck with an arm or leg that’s simply stopped working. They’re asking researchers to batter and overload the neural platforms, to figure out where vulnerabilities can be detected.

DARPA also wants researchers to determine which neural models are the most effective, though they already anticipates that successful prototypes will use “implanted cortical microelectrodes,” to yield the best results. In other words, brain implants that directly communicate with the nervous system. That entails another hurdle: a non-invasive method of monitoring and repairing the devices.


There's more at the link.

I hope DARPA's efforts succeed. I've had occasion to work with amputees who lost limbs in Iraq and Afghanistan during the more recent conflicts there. Artificial limb technology is still in the stone age compared to what DARPA's trying to achieve. If they can come up with a limb that can neurally interface with the brain, interpret what the user wants it to do, and carry out those tasks, it'll be a huge advance, and might give some people back their lives, in a very real sense.

Peter

2 comments:

Jenny said...

Ultimately, I think this technology is going to head for the dustbin not long after it's perfected - if that long. The biotech guys are working on growing real replacements to, which for most people would be an infinitely preferable option.

That said, the neural interface work is going to have some interesting spinoffs.

gary said...

I have been following Video Blogs on the BCI where monkeys were operating robot arms hand to mouth while the Locked-in had to look at a screen to move a cursor to a keyboard. If Darpa has amputees actually working a BCI prosthetic (some so called psychics would pop their toes to simulate spirit tapping) lets see some videos!