Artificial fingertips give robots the touch of a human finger. Image source: DOMINIC PALABISKI
Robots can be programmed to lift a car, or they can assist with some surgery, but if they want to pick up something they haven’t touched before, they tend to fail. Now, engineers have developed an artificial fingertip that enables robots to perceive the texture of an object’s surface like a human fingertip, working with objects of all shapes and sizes. The paper was published in the Royal Society Interface Journal.
Engineers have long tried to make robots as dexterous as humans. Mandayam Srinivasan of University College London, who did not participate in the study, said, “Robots are generally far less tactile than humans. Researchers are bringing the natural and artificial haptic fields closer together, a necessary step to improve robotic haptics. ”
When researchers at the University of Bristol in the United Kingdom began designing artificial fingertips in 2009, they were guided by human skin. The first fingertip is the size of a soda can. By 2018, they turned to 3D printing, making it nearly the size of its tip and all its parts to be the size of an adult’s big toe, and making it easier to create a series of multi-layered structures that approximate human skin.
Recently, researchers incorporated neural networks into fingertips, and this newly developed artificial fingertip is called TacTip. Neural networks help the robot quickly process what it senses and react accordingly — it looks like a real finger.
“Our sense of touch is largely shaped by the mechanical mechanisms of the skin.” Sliman Bensmaia of the University of Chicago in the United States said that the new research is doing a lot of face to face this problem.
When the skin of a human fingertip touches an object, a set of nerve endings deforms and tells the brain what is happening. The nerve then sends a fast signal to help people avoid dropping something, sending a slow signal to convey the shape of the object.
The equivalent signal of TacTip comes from a series of needle-like protrusions beneath the rubber surface that move when they touch the surface of an object. It is like the mane of a comb, hard but bendable. There is also a camera below that can detect when and how the needle is moving. The number of bends of the needle provides a slow signal and the bending speed provides a fast signal. Neural networks translate these signals into fingertip movements, such as gripping or adjusting the angle of the fingertips.
In the new study, Nathan Lepora, an engineer at the University of Bristol, and colleagues tested the artificial fingertip in the same way as assessing human touch. When exposed to corduroy-like materials, they reported that not only could the artificial fingertips detect gaps and protrusions in the material, but the neural signaling patterns they output matched well with those of human fingertips that performed the same test.
In the second project, the Lepora team added more needles and microphones to TacTip, mimicking another set of nerve endings deep inside human skin. As the fingers move across the surface, the nerve endings feel vibrations, enhancing one’s ability to perceive the roughness of the surface. The researchers successfully tested this enhanced fingertip ability to distinguish between 13 fabrics.
However, artificial fingertips are not as sensitive as real fingertips. Lepora notes that humans can detect a gap as narrow as a pencil lead, and TacTip can only notice when it’s twice as wide. But he thinks that once they develop a thinner surface, the resolution will increase.
Lepora says fingertips like TacTip allow robots and prosthetics to handle objects of all shapes and sizes without programming. He’s optimistic about the miniaturization of TacTip because improved 3D printing technology is enabling thinner surfaces that are capable of detecting finer textures and therefore more dexterous. (Source: China Science Daily Wang Fang)
Related paper information:https://doi.org/10.1098/rsif.2021.0822