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-fuelled 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 fibre optics into the design. As Atkeson said himself, "We have a lot of fibre-based elements that hold us together!"
The Soft Robotics and Bionics Lab at CMU is not alone in this crazy endeavour. 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.