Robotics experts built smart blocks that can move and communicate like a swarm of insects, a new development in battery technology could help electric vehicles recharge in the time it takes to get a bag of chips from the gas station, and Russian researchers used artificial intelligence to translate brain waves into images. It doesn’t take a mind reader, though, to enjoy this week’s coolest scientific discoveries.
OK, Blocks: Let’s Get In Formation
What is it? Researchers at MIT’s Computer Science and Artificial Intelligence Laboratory, or CSAIL, have designed a system of “simple, interacting blocks” called M-Blocks that can move individually on their own, climb over one another, jump through the air — and communicate among themselves all the while.
Why does it matter? Like last week’s swarm of drones, researchers envision that this technology could be used in disaster situations, as MIT’s Rachel Gordon explains in an article: “Imagine a burning building where a staircase has disappeared. In the future, you can envision simply throwing M-Blocks on the ground, and watching them build out a temporary staircase for climbing up to the roof, or down to the basement to rescue victims.” The tech is reportedly also inexpensive and simple, compared to other robotics, making it easier to scale.
How does it work? The “M” in M-Blocks stands for motion, magnet and magic but, spoiler alert: It’s not really magic. It’s science! The blocks get momentum from an internal flywheel-and-braking system, and magnets on every side allow them to latch onto other M-Blocks; each face of each block also has a barcode-type identifier that gives the overall system a kind of intelligence, allowing each module to identify and communicate with the others. Algorithms programmed into the blocks help them achieve simple tasks, like assembling into a straight line from a random configuration.
Something About The Way You Move
What is it? A collaboration between scientists at Scotland’s University of Edinburgh and Adobe Research has yielded a way to use deep neural networks to make the movement of 3D animated characters more lifelike.
Why does it matter? The AI will be presented later this month in Australia at SIGGRAPH Asia, a gathering for folks who work in animation, art, gaming and related fields. Researchers behind this gaming- and animation-friendly tech say it makes realistic character movements achievable through “simple control commands.” That’s no small feat, as animators know: A person “walking,” for instance, involves a number of processes, including setting out to walk, placing feet accurately, slowing down toward the end, and so on; that person may also have to realistically interact with objects along the way.
How does it work?”Achieving this in production-ready quality is not straightforward and very time-consuming,” said Sebastian Starke, a PhD candidate at the University of Edinburgh. “Our Neural State Machine instead learns the motion and required state transitions directly from the scene geometry and a given goal action.” The team trained its AI on motion capture data; it analyzes a character’s previous poses, and the geometry of a given scene, to infer next steps. Starke said, “Our method is able to produce multiple different types of motions and actions in high quality from a single network.”
Getting Electric Vehicles To Accept The Charges
In a battery, ions flows from the cathode to the anode, resulting in a positive energy charge for the unit. Image and caption credits: Chao-Yang Wang Lab, Pennsylvania State University.
What is it? A team of Penn State engineers came up with a way to charge electrical vehicles in 10 minutes with enough juice to allow them to travel 200 or 300 miles; crucially, the technique doesn’t degrade their batteries.
Why does it matter?“Fast charging is the key to enabling widespread introduction of electric vehicles,” said Chao-Yang Wang, an engineering professor and director of Penn State’s Electrochemical Engine Center. One obstacle to broad EV adoption is range anxiety— drivers’ concerns about being able to travel far enough on a charge, easily find a charging station once their battery is low, and not have to wait around forever till they can drive again. As Wang and colleagues note in a paper in Joule, the Department of Energy (which supported this research) has set a goal for developing extrafast technology that can charge an EV in about as much time as it takes today’s drivers to fuel up, use the bathroom, and grab a snack. It’s “essential for adoption of electric vehicles because it solves the range anxiety problem,” Wang added.
How does it work? The heart of EV technology, lithium-ion batteries, have a kind of sweet spot when it comes to the temperature at which they recharge — too cold or too hot and the batteries degrade. Wang and his colleagues found that they could charge batteries at 140 degrees Fahrenheit in 10 minutes if the batteries were then rapidly cooled to ambient temperatures — without damage to the battery. They designed a battery that heats itself using thin nickel foil, then takes advantage of the car’s cooling system to chill out after it’s finished charging. In this way, Wang says, the battery can withstand 2,500 charging cycles — or about 500,000 miles of travel.
What is it? A team of Russian researchers from the company Neurobotics and the Moscow Institute of Physics and Technology designed a way to visualize a person’s brain activity by reconstructing it from neural signals. Their tools? Artificial neural networks and data from electroencephalograms, or EEGs.
Why does it matter? Besides just being wildly cool — this program can, in essence, create images that reflect what it sees in somebody’s brain — this kind of technology could help stroke patients control an exoskeleton arm, for instance, or it could help people who’d been paralyzed control an electric wheelchair. The paper describing it is available on the preprint server bioRxiv; as co-author Grigory Rashkov explains, it’s also a step forward for how neuroscientists can use EEGs. “Researchers used to think that studying brain processes via EEG is like figuring out the internal structure of a steam engine by analyzing the smoke left behind by a steam train,” Rashkov said. “We did not expect that it contains sufficient information to even partially reconstruct an image observed by a person. Yet it turned out to be quite possible.”
How does it work? The team showed subjects clips of YouTube videos sorted into five general categories: abstract shapes, waterfalls, etc. Analyzing EEG data taken from subjects while they watched, researchers were able to match brain wave patterns to specific categories. They then designed two neural networks, as MIPT explains in a release: “one for generating random category-specific images from ‘noise,’ and another for generating similar ‘noise’ from EEG.” Working in tandem, these networks were able to render images based on EEG signals from test subjects; in subsequent tests, subjects were shown new videos from the same categories while wired up to the EEG, which passed their brain waves along to the neural networks. The AI “generated convincing images that could be easily categorized in 90% of the cases.”
Top image credit: Getty Images.
Is There No Hope For The Suture?
What is it? Doctors might finally get access to a technology that crafters have known about forever: double-sided tape. A team from MIT has developed a double-sided adhesive that could be used to rapidly seal tissue together, and even hold medical implants in place.
Why does it matter? The current surgical state of the art is suturing, which has some drawbacks, as MIT engineering professor Xuanhe Zhao explains: “There are over 230 million major surgeries all around the world per year, and many of them require sutures to close the wound, which can actually cause stress on the tissues and can cause infections, pain and scars. We are proposing a fundamentally different approach to sealing tissue.” Zhao is the senior author of a new study in Nature.
How does it work? Engineers trying to devise an adhesive system for tissue face the same challengers as glue makers everywhere: It doesn’t stick as well when wet. To solve this, the MIT team drew its inspiration from spiders. To capture prey, spiders rely on “glue” that uses charged polysaccharides to absorb water off their targets almost instantly, allowing it to stick. The researchers’ version uses a two-step process: Their adhesive uses polyacrylic acid — an absorbent material used in diapers — to rapidly absorb water from tissue and form weak hydrogen bonds; that holds the tape in place long enough for another group of compounds, called NHS esters, to solidify a stronger bond. They tested their material successfully on different types of pig tissue, including skin, intestine, stomach, and liver.