Instinctual grip exclusive to conscious beings can now be coded into AI

5 minutes ago

Robot grip is becoming more human all the time. (Lilian Hsiao)

Algorithms are required to give traits to robots — they're computers that lack intuition, after all — and newly released formulas represent a specific type of friction intrinsic to human grip that takes into account fingerprints, sweat and dexterity.

Published Thursday in Nature Materials, these mathematical equations could be applied to such algorithms and are the first to iterate the precise laws of a little-studied form of friction called elastohydrodynamic lubrication, or EHL, friction. It dictates the way that patterns on soft surfaces interact with thin films of wetness, such as when fingerprints on sweaty hands are rubbed together.

"When we want to touch objects — for example, when we touch our phones — typically, we slide our finger perpendicular to the screen," said corresponding author Lilian Hsiao, an assistant professor of chemical and biomolecular engineering at North Carolina State University. "That's actually representative of these patterns being dragged across."

Humans intuitively adjust their grip when objects begin slipping out of their hands, easily navigating the duality of moisture and fingerprint ridges. That's because from a very young age, humans are taught to not drop things; the brain involuntarily communicates with mechanical receptors in the skin to manage friction accordingly.

For instance, if a soapy dish starts to slide away while being washed, a human would hold it a little harder. Because that type of dexterity is intrinsic for conscious beings, there was never really a reason to explore it — and then came artificial intelligence.

Robots do not possess such natural instinct, so they struggle in wet environments.

"There hasn't been an easy way to translate this to synthetic systems; for example, how should a robot adjust forces?" Hsiao explained to The Academic Times. "If we don't understand it — because we haven't studied it in the human system — then the robots can't do it, because we don't know how to program it."

This gap in understanding the laws of EHL friction is what motivated the team to solidify its principles. 

The team's framework is not exactly a new law of physics, but rather a modified version of the well-known principles of fluid flow and friction attributed to Irish mathematician and physicist Osbourne Reynolds

"What we've done here is taken this further and combined that with solid mechanics, in which they care a lot about compressional surfaces and defamation of surfaces when you have pressure," Hsiao explained.

She and her colleagues realized that patterns — solids — have a huge impact on friction between wet or moist surfaces. That's why they incorporated friction principles from solid mechanics into the original fluid equations.

"This was very important for patterns, and it was also very important for soft things because if you press them, they deform," she said. "So, we took these two principles — these two fields, really — and combined the governing equations."

Anyone can harness these new laws defining EHL to adjust dimension, softness, patterns or size of a material. And that, Hsiao says, is what will transform technology.

Put to robotics, the formulas can lead to automated algorithms that almost intuitively adjust moisture and pressure in accordance with a pattern on a robotic arm — just the way humans can with their naturally oily hands and fingerprints. 

"We're now designing new types of materials where the amount of liquid can be changed on demand for robotics, and actually, we are inspired by animals — amphibians and frogs — because they also have patterns on their skin and they can change the amount of liquids between themselves and other objects," Hsiao said. "We can then basically control the friction at will, which hasn't been done before in other areas at all."

But the consequences of the new formulas stretch beyond artificial intelligence. Now that characteristics of the friction have been defined, the framework can potentially be harnessed to manipulate or enhance human sensation, too.

"We were very surprised when we found out that we could apply this [framework] to actual human skin, which is very complex," Hsiao said. "It's a little dirty. It's not perfect. Everybody has a different skin; everybody has different finger patterns."

Despite being a savior for social connection during COVID-19, digital platforms such as Zoom cannot offer the thing that humans crave: touch. That's because it's virtually impossible to design sensation that can be shared via wireless connection. 

"We don't know how to design a handshake," Hsiao said. "If we understand how to design materials that can give you different types of frictional feelings, perhaps we can translate it into electrical signals."

She added, "If you're in New York, I'm in Raleigh, maybe we can shake hands at some point."

Another application could be to prosthetics, though Hsiao said all of these human-centric implications are likely to happen in the next five to 10 years.

"Right now, we just kind of fit a prosthetic that doesn't provide that much sensation," she said. "But what you can imagine is having a skin that — it doesn't matter if it's dirty — and this person can still feel the same way that they would feel if their skin were to be healthy."

The study, "Elastohydrodynamic friction of robotic and human fingers on soft micropatterned substrates," published April 29 in Nature Materials, was authored by Christopher M. Serfass, Catherine N. Hill, Lilian C. Hsiao and Yunhu Peng, North Carolina State University; Anzu Kawazoe, Yitian Shao and Yon Visell, University of California, Santa Barbara; and Kenneth Gutierrez and Veronica J. Santos, University of California, Los Angeles.

We use cookies to improve your experience on our site and to show you relevant advertising.