Biologists transferred memories from one animal to another, scientists created a 3D printer for skin, and an already agile humanoid robot just got even more limber. This is the fast lane for science news, folks, and the future is roaring forward.
What is it? MIT spinoff Boston Dynamics released a video of the latest trick learned by its unsettlingly humanoid robot, Atlas — this one features it taking a casual jog through the Massachusetts suburbs, proving he can navigate autonomously on uneven outdoor terrain.
Why does it matter? As we’ve reported previously, Atlas is already capable of walking, jumping, doing backflips, avoiding “mean pranks” played by humans — it’s quickly evolving to mimic many human functions. Boston Dynamics — which you may remember from such hits as Big Dog, “the first advanced rough-terrain robot” — is making a great effort to stay ahead of the pack. Atlas could be terrific at dangerous tasks like search and rescue.
How does it work? Atlas’ hardware, according to Boston Dynamics, “takes advantage of 3D printing to save weight and space, resulting in a remarkably compact robot with high strength-to-weight ratio and a dramatically large workspace.” Its dimensions are that of an adult human — 180 pounds, 5 feet 9 inches. The battery-powered machine operates by flexing 28 joints, relying on lots of hydraulic actuators and using LIDAR for navigation and stereo vision.
What is it? A team from MIT’s Computer Science and Artificial Intelligence Laboratory, or CSAIL, has developed a technology that allows self-driving cars to safely navigate rural roads that haven’t previously been mapped in 3D.
Why does it matter? Most self-driving cars stick to urban streets, where companies have developed comprehensive 3D maps of the terrain — so the vehicles know what they’re getting in terms of ramps, curbs, lanes and so on. That’s not the case for millions of miles of less-trafficked roads in rural areas, where 3D-mapping companies haven’t collected as much data. CSAIL’s technology, called MapLite, helps self-driving cars navigate roads they hadn’t been on before, and for which 3D maps aren’t available.
How does it work? According to MIT, MapLite “combines simple GPS data that you’d find on Google Maps with a series of sensors that observe the road conditions.” The team tried it out in the country around Devens, Massachusetts, finding that this combination enabled its self-driving cars to “reliably detect the road more than 100 feet in advance.” CSAIL grad student and lead author Teddy Ort said, “The reason this kind of ‘map-less’ approach hasn’t really been done before is because it is generally much harder to reach the same accuracy and reliability as with detailed maps. A system like this that can navigate just with on-board sensors shows the potential of self-driving cars being able to actually handle roads beyond the small number that tech companies have mapped.”
Top image: MapLite uses perception sensors to plan a safe path, including LIDAR to determine the approximate location of the edges of the road. Image credit: CSAIL. Caption credit: MIT News.
What is it? Using machine learning, scientists trained a computer program to navigate its way through space like an animal would — and it surprised them by “spontaneously generating” patterns of activity that mimic the brain as it maps its environment.
Why does it matter? In 2014, a group of researchers won the Nobel Prize for their work on grid cells — neurons that help many species navigate, like a kind of neural GPS, by firing in a particular pattern as animals explore their environment. But much about grid cells remains unknown. It’s been hypothesized, for instance, that they assist in vector-based navigation, helping animals get from one point to another as the crow flies — even if they’ve never been to their destination before. As reported in a new paper in Nature, a team of researchers from the London-based Google company DeepMind and the University College London wanted to use AI to test the theory that grid cells work in support of vector-based navigation.
How does it work? The team programmed a computer to simulate a rat running through a maze. According to Nature, “Grid cells have been shown in experiments with real rats to be fundamental to how an animal tracks its own position in space.” The computer rat surprised everybody by spontaneously creating hexagonal-shaped patterns of activity that are characteristic of the functioning of grid cells. Edvard Moser, one of the aforementioned Nobel winners, observed, “It is striking that the computer model, coming from a totally different perspective, ended up with the grid pattern we know from biology.”
What is it? Researchers at the University of Toronto have developed a handheld device that essentially 3D-prints skin using biomaterials — representing a potential big leap forward for doctors treating deep burns and other wounds.
Why does it matter? One popular method for burn treatment, skin grafting, can require a lot of donor skin if the wound is very large — and if enough can’t be harvested, doctors are forced to leave portions of deep burns uncovered, leading to problems healing. Meanwhile, as Toronto professor Axel Guenther told Engadget, “Most current 3D bioprinters are bulky, work at low speeds, are expensive and are incompatible with clinical application,” making them impractical for routine use. The device Guenther’s team came up with, by contrast, weighs less than a kilogram and takes just minutes to form, deposit and set tissue, making it ideal for surgeries.
How does it work? Using a “microfluidic cartridge,” the device deposits biomaterials like fibrin — a protein involved in blood clotting — and connective tissue like collagen to form sheets “of consistent thickness, width and composition” on the skin, according to a new paper in the journal Lab on a Chip. The team has successfully tested the tech on rodents and pigs, and plans more in vivo experimentation before beginning testing on humans.
What is it? When biologists at UCLA injected RNA from one marine snail into another, they were able to transfer a memory between the snails. In short, they used ribonucleic acid to implant a fake memory.
Why does it matter? Beyond being grist for a great “Black Mirror” episode, the technique could be used in the “not-too-distant future to ameliorate the effects of Alzheimer’s disease or post-traumatic stress disorder,” said UCLA biologist David Glanzman. It could also restore memories that have been lost, or help alleviate the effects of traumatic memories. The results of the team’s work were published May 14 in eNeuro, the online journal of the Society of Neuroscience.
How does it work? Researchers delivered a series of mild electric shocks into the tails of a group of snails, and then another series 24 hours later. Relative to a control group, the snails that received the shocks displayed an enhanced “defensive withdrawal reflex,” as a release from UCLA put it — the shocked snails had become more sensitive. The team then extracted ribonucleic acid, an important cellular messenger, from the shocked snails and injected it into a group of unshocked snails — and found that when subsequently stimulated, the new group displayed a defensive response as if they themselves had gotten shocked. “It’s as though we transferred the memory,” Glanzman said. He noted that his findings challenge a commonly held belief that memory is stored in the synapses; to Glanzman, the snail experiment suggests that it is contained, instead, in the nucleus of neurons. And apparently in the snail in the other room, too.