Tech

AI-powered artificial fingers give robots a near-human touch

slash dot reader science habit share this article. Science magazine:

Robots can be programmed to lift cars and perform some surgeries, but often fail miserably when picking up objects they haven’t touched before, such as eggs. Now engineers Find an artificial finger that goes beyond that limit. These advances allow machines to detect the texture of these surfaces just like human fingertips….

[W]Researchers at the University of Bristol began designing artificial fingers in 2009, using human skin as a guide. Their first fingertips, assembled by hand, were the size of a can of soda. By 2018, they had switched to 3D printing. This made the tip and all components the size of an adult big toe and made it easier to create a series of layers close to the multi-layered structure of human skin. More recently, scientists have integrated neural networks at their fingertips called TacTips. Neural networks help the robot quickly process what it detects and react accordingly. They look like real fingers.

At our fingertips, the layer of nerve endings deforms when the skin comes into contact with an object, telling the brain what’s going on. These nerves send “fast” signals to avoid dropping objects and “slow” signals to convey the shape of objects. TacTip’s equivalent signal comes from an array of pin-shaped protrusions beneath a layer of rubberized surface that moves when you touch the surface. The pins of the array are like comb teeth that are stiff but bendable. Beneath that arrangement is a camera that detects when and how the pins are moving, among other things. The degree of bending of the pin gives a slow signal and the speed of the bending gives a fast signal. Neural networks transform these signals into fingertip motions, for example to grip more tightly or adjust the angle of the fingertip….

In the second project, Lepora’s team added more pins and microphones to the TacTip. Microphones mimic another set of nerve endings deep in the skin that sense the vibrations we feel as we move our fingers across the surface. These nerve endings improve your ability to feel how rough the surface is. that much So was the mic. When the researchers tested the improved fingertip’s ability to discriminate between 13 fabrics.
This article states that the output of the artificial finger in the test was “closely Matches neural signal patterns in human fingertips They are undergoing the same tests.”

AI-powered artificial fingers give robots a near-human touch

Source link AI-powered artificial fingers give robots a near-human touch

Back to top button