The computer—or more accurately, the Turing Machine—changed the world with a groundbreaking idea: Any piece of information could be coded in 0s and 1s. And so theoretically, any question could be answered by sorting these numbers through an automated process. Even today, in the era of microprocessors and 4G Internet, it’s a rendition of these 0s and 1s that apply Instagram filters, power Google’s predictive search, or render headshots in Call of Duty.
Working under a grant from DARPA, Neil Gershenfeld, head of MIT’s Center for Bits and Atoms, along with graduate students Ara Knaian and Kenneth Cheun, have flipped this idea on its head. Rather than turning real ideas into binary code, they’re turning binary code into real ideas.
They’ve created the robotic manifestation of raw digital data called the Milli-Motein. At just a few centimeters long, this metal caterpillar is a bit unimpressive to look at, but so is binary code. In reality, it’s a one-dimensional robot, a segmented strand of 0s and 1s that’s theoretically capable of bending into any shape or structure you can imagine. Coffee cups. Airplane turbines. Anything. “You give us a shape, we give you a code to fold it,” Gershenfeld tells me.
I promise, the idea is nowhere near as unapproachable as it sounds. So you’re looking at a worm, right? Each segment of that worm has one of two options, to turn left or turn right (that’s the 0 or the 1 value). No matter what shape you want to make, any one part of this worm only needs to know left or right.
The question then becomes, why isn’t the Milli-Motein just one giant zig-zag? That’s because these left and right turns are made in relationship to the position of the previous segment in the chain. Consider a compass. If you turn left when you’re heading north, you actually head west. And if you turn left again, then you’re heading south. So while the robot may be one-dimensional, it can be bent into three dimensions. And while the robot may be folding incredibly complex shapes, it’s infinitely scalable, as each segment needs only enough processing power to handle a single 1 or 0—a mere bit of information—along with a motor to turn it left or right.
On one hand, the Milli-Motein is a radical idea. Any one Milli-Motein circuit could reshape itself to become a smart component of a larger machine, like a turbine, a wheel, or a fender. But on the other hand, its innovation is based on a process that’s several billion years old, entrenched in every single cell in our bodies: ribosomes. Ribosomes are proteins that make proteins. They use a process called elongation to build proteins from one-dimensional chains. These chains, via the intricate miracle of protein folding, become the molecular machines in our bodies that sense light in our eyes or move muscles in our arms. It’s this 1:1 parallel that allows Gershenfeld to confidently call the Milli-Motein “a mechanical protein, or programmable matter.” I like the term “physical software,” too.
Truth be told, MIT has been pursuing similar programmable matter for some time now. Previous explorations have examined the use of stacking, sticking blocks—or robotic pebbles—to build complex shapes.
But thanks to a few mechanical breakthroughs, the new chain design may have certain long-term advantages. For one, the Milli-Motein chain is a single, long circuit. That means it could be mass manufactured much in roll-to-roll fashion, as if you’re printing one long newspaper to be chopped at the end of printing.
The single circuit design also means that you only need to send power and information through one end of the structure, which will spread along the spine to each node along the line much like a train making station-by-station stops. Compare that one-line distribution to negotiating power and data across countless discrete blocks, which would have all the complexity and randomness of a city’s taxi system.
Yet if there are so many advantages to this protein-like structure, why not just create it in the first place? To build a robot of this size, the lab needed a breakthrough: an entirely new type of motor.
“Conventional electric motors wouldn’t work for two reasons,” Gershenfeld explains. “They need power to hold their position. (What you’d like is to turn the robot into the shape you want and have it stay there.) And for a conventional electric motor, moving slowly is inefficient. If you stall the motor, all the energy goes into heat.”
Electric motors turn at fast speeds—just imagine the high-pitched squeal of the average power drill. Generally, fast-running motors can just use gears to slow down their rotation, but gears couldn’t possibly be squeezed into the Milli-Motein’s frame. Instead, what the lab created was a magnetic motor that’s capable of moving with very little power and holding a position passively (through magnetism). Electro-magnetism works its way in steps around a ring, with each part of the rotation requiring but a sip of power to cascade its polarity like a domino. Somewhat ironically, despite the scope of MIT’s programmable matter, it’s this motor that Gershenfeld believes will be commercialized first.
“We’re working with aerospace industry,” he explains. “Instead of hydraulics in planes, there’s a move to have lightweight electric motors. Our motor could work for anything involving large forces moving in slow motions.”
At the moment, the Milli-Motein is the smallest reconfigurable robot that’s ever been made. Gershenfeld’s team has created a larger version—as big as a human—and is in the process of developing a much smaller, molecular version, that operates on the nanoscale. Needless to say, a nanomotor uses a totally different process to drive each joint, but the concept is fundamentally the same. That’s why all these robots are generically labeled "moteins" by MIT.
Each of these robots could serve a different size of problem—one operating inside your house, the other operating inside your body. But the foundational idea is equally revolutionary across the board, and the design repercussions of the Milli-Motein can’t be overstated, even by Gershenfeld himself.
“The really interesting frontier isn’t just bending a piece of plastic, it’s making an object where the material itself can change shape,” Gershenfeld says. “This robot is an important step in the development of shape changing materials.”