Clearing The Air
What is it? Researchers affiliated with Graphene Flagship, a collaboration funded by the European Union and coordinated by Chalmers University of Technology in Sweden, announced they’d designed a “smog-eating graphene composite” that could be applied to concrete to combat pollution.
Why does it matter? Specifically, the technology targets a group of pollutants called nitrogen oxides, or NOX, which come from automotive exhaust and the burning of fossil fuels. Those pollutants react with sunlight and other chemicals to form smog. One form of NOX is nitrogen dioxide, which is linked to asthma in children, and lung cancer and cardiovascular problems, according to the American Lung Association.
How does it work? Graphene Flagship research partners — involving institutions in Israel, Italy, the Netherlands and elsewhere — designed a photocatalyst, which uses light to generate a chemical reaction, made of a composite of graphene and titania, a type of titanium oxide nanoparticles. “When titania is exposed to sunlight,” according to a news release from Graphene Flagship, “it degrades nitrogen oxides ... oxidizing them into inert or harmless products.” Researchers discovered that the addition of graphene helped the photocatalyst degrade up to 70% more pollutants than titania alone. The composite material could be applied as a coating on streets or buildings, researchers said, where rain or wind can wash away the pollutants after they’ve been rendered harmless. The technology is described further in Nanoscale.
What is it? Artificial intelligence can be humble and curious — but can it be surprised? That’s the idea behind a new AI model designed by researchers at MIT, who are trying to see whether computers can understand “intuitive physics” like very young humans do.
Why does it matter? Early in their development, infants gain an understanding of physical space that involves the element of surprise when, for instance, a moving object does something unexpected. By building the same sorts of expectations into their model, the MIT researchers are hoping both to design smarter AI and to get a better understanding of human cognitive development. “By the time infants are three months old, they have some notion that objects don’t wink in and out of existence, and can’t move through each other or teleport,” said MIT research scientist Kevin A. Smith, first author of a paper presented at the Conference on Neural Information Processing Systems in Vancouver, British Columbia. “We wanted to capture and formalize that knowledge to build infant cognition into artificial intelligence agents. We’re now getting near humanlike in the way models can pick apart basic implausible or plausible scenes.”
How does it work? Smith and colleagues designed a model, called ADEPT, that’s presented with video of a set of objects moving around a scene. Having a general understanding of how the objects should behave given their underlying physical properties, the model signals a sense of “surprise” when they do otherwise — like vanishing altogether, or disappearing and then reappearing elsewhere in the scene. Like babies, the model doesn’t get bogged down in the details, Smith said: “It doesn’t matter if an object is rectangle or circle, or if it’s a truck or a duck. ADEPT just sees there’s an object with some position, moving in a certain way, to make predictions. Similarly, young infants also don’t seem to care much about some properties like shape when making physical predictions.”
What is it? A collaboration led by researchers at the University of Bath in the United Kingdom developed the first “artificial neuron” on a silicon chip, which they hope could someday lead to cures for chronic diseases such as Alzheimer’s and heart failure.
Why does it matter? One cause of heart failure, for instance, is malfunctioning neurons in the base of the brain that send the wrong signals to the heart, causing it to pump improperly. To treat these and other life-threatening conditions, medical researchers have sought for decades to synthesize neurons, a devilishly complicated task, said University of Bath physics professor and project leader Alain Nogaret: “Until now, neurons have been like black boxes, but we have managed to open the black box and peer inside. Our work is paradigm-changing because it provides a robust method to reproduce the electrical properties of real neurons in minute detail.” The synthetic neurons also require just a billionth the power of a microprocessor, Nogaret said, making them well-suited for medical implants.
How does it work? Nogaret and colleagues first modeled and created equations to describe how biological neurons respond to electrical stimuli in the nervous system, then designed silicon chips that accurately reflected biological ion channels. “Our approach combines several breakthroughs,” Nogaret said. “We can very accurately estimate the precise parameters that control any neuron's behavior with high certainty. We have created physical models of the hardware and demonstrated its ability to successfully mimic the behavior of real living neurons. Our third breakthrough is the versatility of our model, which allows for the inclusion of different types and functions of a range of complex mammalian neurons.” The development is described further in Nature Communications.
What is it? Can light pass through trees? Not quite, but researchers at VTT Technical Research Centre of Finland did just come up with an optical fiber made of wood-based cellulose.
Why does it matter? Using cellulose in optical fiber is attractive for a couple of reasons, according to a VTT news release: “Cellulose effectively absorbs and releases water, which can be measured by the change in the attenuation of light transmitted in the fiber.” And cellulose-based optical fibers could be used in sensors that would benefit from the ability to biodegrade. Beyond that, said VTT senior scientist Hannes Orelma, the R&D is still in an initial phase, and not all applications are known yet.
How does it work? The researchers used ionic solvents to modify cellulose to carry light through the center of an optical fiber; it’s coated with a cladding of cellulose acetate that reflects light back into the fiber’s core. The technology is described further in the journal Cellulose.
Teaching An Artificial Dog New Tricks
What is it? Researchers at Finland’s Tampere University and Aalto University applied an old behavioral trick — Pavlovian conditioning — to a new subject: artificial materials. They used heat and light to “train” materials to adopt new behaviors.
Why does it matter? As a Tampere press release explains, this work is unique because it’s interdisciplinary, bringing together “fundamental research in materials chemistry and concepts from psychology.” The techniques the researchers are studying could someday be used to create novel coatings, for instance, that adjust their own properties — like color — in response to various stimuli. “Our concept of learning can also be applied to other areas, although at the moment conditioning can only be achieved through interaction between light and heat,” said Tampere’s Arri Priimägi.
How does it work? In one experiment, described in Nature Communications, Priimägi and colleagues created a gel from a light-responsive substance and tiny gold nanoparticles, which are capable of self-assembling into chains. They “conditioned” the material by heating it while exposing it to a certain wavelength of light. Afterward, when it was exposed to the same wavelength of light without being heated, it melted nonetheless, having “learned” to respond to the light alone. To drive the point home, the researchers kept the material in a dog-shaped container. When exposed to a particular wavelength of light, then, the gel would melt and the animal would appear to drool — the same sort of conditioned response Ivan Pavlov provoked in actual dogs more than a century ago."