This Bionic Hand Uses AI to Grab Things Automatically
It feels more like a real hand.
Image: Newcastle University
Most moveable prosthetic limbs require some human control, transmitted through devices like implanted myoelectric sensors, to react to their environment—translating the movement to a cyborg appendage.
What if the hand could see for itself, instead?
Biomedical engineers at Newcastle University in the UK designed the hand to "see," with a camera attached to the knuckles. Using neural networks, the basis for artificial intelligence, to recognize the object in front of it, the hand reacts to grab what's nearby within milliseconds. That's around 10 times faster than current limbs on the market, the researchers found.
This is a step on the way to much more advanced prosthetics. They're hoping to develop a limb that connects electrodes to the nerve endings in the arm, for a seamless communication from one's brain to the bionic fingers. For the seeing-hand, they've started trials on a small group of amputees.
Dr. Kianoush Nazarpour, a lead author on the study published in the Journal of Neural Engineering, told me that the most advanced object that this hand has attempted to grab so far is a pair of scissors. If it goes too roughly grab a wiggling kitten (for example), or to make a too-firm handshake, the user can override the bionic hand, reset the decision and try again—assuming everyone survived.
Subscribe to Science Solved It , Motherboard's new show about the greatest mysteries that were solved by science.