Stanton M.,Carnegie Mellon University |
Sheng Y.,Carnegie Mellon University |
Wicke M.,Otherlab |
Perazzi F.,Carnegie Mellon University |
And 3 more authors.
ACM Transactions on Graphics | Year: 2013
This paper extends Galerkin projection to a large class of nonpolynomial functions typically encountered in graphics. We demonstrate the broad applicability of our approach by applying it to two strikingly different problems: fluid simulation and radiosity rendering, both using deforming meshes. Standard Galerkin projection cannot efficiently approximate these phenomena. Our approach, by contrast, enables the compact representation and approximation of these complex non-polynomial systems, including quotients and roots of polynomials. We rely on representing each function to be model-reduced as a composition of tensor products, matrix inversions, and matrix roots. Once a function has been represented in this form, it can be easily model-reduced, and its reduced form can be evaluated with time and memory costs dependent only on the dimension of the reduced space. Copyright © ACM 2013.
News Article | April 15, 2015
Last week, an artificially intelligent robot scared me to death . The next day, I travelled to Carnegie Mellon University where I met a lab full of robots designed to do the exact opposite. Big, soft, and inflatable, these robots are Disney characters in real life. Your grandma’s going to love them. That’s the idea. I first encountered CMU’s soft robots in a crowded lab run by a towering man who looks like a cross between Jerry Garcia and Baymax, the benevolent inflatable robot from Big Hero 6. This is appropriate because this man, robotics professor Dr. Chris Atkeson, helped build the soft robot that inspired Disney’s portrayal of Baymax in Big Hero 6. One of his primary areas of interest is in humanoid robots and human-aware environments. The burgeoning field of soft robotics is exactly what it sounds like: a new method of building robots not out of hard, dangerous metal but rather out of soft, safe materials. One particularly promising method for building soft robots involves inflatable elements that can vary in firmness. In the far future, this technology could resemble the touch and feel of human flesh. For now, soft robotics look pretty much like balloons. Pretty cute, huh? That demo happened at CES in 2011, around the time that Big Hero 6 director Don Hall found the inspiration for a big, huggable Baymax. If we’re ever going to be friends with robots, we probably want them to look more like Baymax and less like the Terminator without skin. As such, the great hope for soft robotics is a future in which agile robots can interact with humans safely and without intimidation. It’s a noble and promising goal. Atkeson talked to me about how friendly androids could help the elderly live longer in their own homes. I imagined a Disneyesque scenario where plush robots watched over children on playgrounds and helped the disabled across the street. These scenes were the polar opposite of the robot-fueled violence I’d seen on screen just a couple days ago. I liked this vision for robotics. “A big problem with traditional robotics is the safety issue,” Atkesonsaid in the Soft Robotics and Bionics Lab at CMU. “That’s holding us back.” He continued, “Soft robotics is based on the technology we use to make clothes and toys. In some sense, we’ve just begun.” Along with the Soft Machines Lab—where mechanical engineering professor Carmel Majidi and his team make squishy, flexible circuits—Atkeson’s lab is making progress building more robust soft robots. A graduate student gave me a demo of a very robotic-looking arm that was completely powered by air. The muscles and tendons were actually rubber-coated hoses that could flex and expand based in varying amounts of air pressure. The sound of it is mildly horrifying, but it’s entirely safe. Furthermore, the whole thing is eventually supposed to stay covered in a soft, sensor-laden skin. It’s all utterly mind-boggling when you step back and think about it. Probably due to movies and comic books, I’ve always imagine the future would be filled with boxy grey talk bots that moved in a stiff calculated fashion. But the more Atkeson told me about how everything worked, it was obvious that this makes little sense. Not only do heavy materials gobble up battery life, but the mechanics of hard metal robots make agility a tough challenge. Just imagine a robot that doesn’t know how strong it is going to give grandma a hug. Soft robotics are air-powered and designed not to crush anything. As the name of his lab implies, Atkeson and his team take cues from nature. Humans and most other animals are soft and squishy, and that seems to be working out pretty well. Why not aim for that design? “It’s a crazy idea at all levels,” Atkeson told me. “We have to demonstrate that we can make robust, useful robots.” Out of fabric and air. It is absolutely crazy. But the more this Baymax-shaped professor tells me about composite elastimers and liquid metal alloys, the more I’m picking up what he’s putting down. After all, like his colleagues building snake-bots down the hall , Atkeson is taking inspiration from nature and building bionic machines that simply work better for more purposes. One ongoing challenge involves creating touch sensitive skin. Another strives to avoid requiring a column of wires going down the spine of these robots. Down the line, the professor hopes to integrate fiber optics into the design. As Atkeson said himself, “We have a lot of fiber-based elements that hold us together!” The Soft Robotics and Bionics Lab at CMU is not alone in this crazy endeavor. The New York Times recently profiled Otherlab, a research company in San Francisco that’s building everything from inflatable exoskeletons to plush factory pickers with arms nimble enough to draw pictures. “Every problem in mechanical engineering has been addressed with more weight, more power and more stiffness,” Saul Griffith, cofounder and chief executive of Otherlab, told the Times. “But nature—the real world—is squiggly.” Soft robotics is a long game. Rather than attempting to compete with the small but impactful existing robotics market, Griffith and his team are looking half a century into the future, when we won’t just need robots to do things, we’ll need them to do things alongside human beings, and without scaring anybody. It’s going to take some time to figure out that fine balance. In Griffith’s words: “If you’re going to make robots like you see in the movies, you have to change the game. We’re trying to look at what manufacturing will be in 50 years.” Others have gotten hip to the soft robotics movement. iRobot, the maker of the Roomba vacuum, also now has a soft robotics initiative. Harvard researchers recently released a soft robotics toolkit to make it easier to design soft robots. A peer-reviewed soft robotics journal opened up shop last year. And meanwhile—perhaps obviously—DARPA is funding millions of dollars worth of research and development and surely exploring the military applications of this technology. Soft robotics on the battlefield is missing the point, though. Grey-beareded and imminently cheerful, Atkeson frames the technology as approachable and familiar. He describes his purpose in plain English. “We want to develop robots that help old people live in their homes longer,” he said to me smiling, as we wrapped up my lab visit last week. And if you and I are lucky, maybe we’ll live long enough to take advantage of big, soft, futuristic friends like Baymax.
News Article | February 25, 2013
Creating compelling narratives and telling solid stories can be a game-changer for early stage energy technologies — for developing products, for pitching investors, and for gaining customers and partners. At the fourth annual ARPA-E Summit on Monday around 4:30PM EST (1:30PM PST) we’ll be live streaming a discussion between Otherlab’s Saul Griffith, IDEO’s Dave Blakely, and myself, about the power of narratives for energy tech. Don’t miss this! It’s one of the only live, free online events for the show. ARPA-E is a program created by the Department of Energy to give small grants to early-stage, high-risk energy technologies that can be game-changers. Here’s to moonshots! They need some powerful stories. Watch to find out why. (If we’re running a few minutes late, be patient, we’ll start soon).
News Article | November 23, 2011
Otherlab's new 15-foot inflatable robot looks like something we created in Spore a few years back, but instead of evolving into a space-faring civilization, the pneumatic Ant-Roach will spend its time on Earth at the bidding of clever humans. The creators devised the robot with the goal of demonstrating the carrying capacity and strength of inflatable structures, and it looks like they've succeeded — the Ant-Roach weighs less than 70 pounds (not including the external air compressor) and Hizook speculates that it could probably support up to 1,000 pounds. The robot's "muscles" are fabric actuators which create motion by contracting into specific shapes, and the whole system is directed via wireless laptop. The Ant-Roach might not appeal to our vanity — it can't read our facial expressions or pour us a drink — but it's a really impressive engineering feat nonetheless. Be sure to check out the robot in action below.
Agency: National Aeronautics and Space Administration | Branch: | Program: SBIR | Phase: Phase II | Award Amount: 1.50M | Year: 2015
We have invented a new class of robotics, called `Pneubotics', that rival current manipulators in payload and reach at 1/10th the weight. Our technology leverages insights into lightweight materials and mass manufacturing to create robots that derive power, structure, and movement from pressurized air. As a result, drive trains, motors, bearings, shafts, sliding surfaces, and excess structural material are eliminated, leading to robots with extremely high strength to weight ratios, inherently human safe operation, and high degrees of freedom at low part count. This transformative new technology has the potential to enable the widespread use of automated handling of material and equipment on missions in low Earth orbit and beyond. The compliant nature of these robotic systems allows them to robustly grasp arbitrarily shaped objects and makes them ideal for operating around sensitive equipment and materials or cooperatively with humans. Similarly, due to their fluidic architecture they can be deflated and stowed for efficient transport. The work described in this phase II SBIR proposal would integrate the component development and analysis performed in Phase I to build and test a full prototype manipulation system. By incorporating optical, internal, and tactile sensors and multi-level controls that take advantage of the unique characteristics of the manipulator and seek out appropriate contact to guide motion rather than avoiding it. By testing the entire prototype system in the field we will demonstrate operation in the ground environment and learn valuable lessons for IVA and EVA applications.