If the movies have taught us anything, it’s that the future isn’t just about robots–it’s about robots that can heal, adapt, and change their entire appearance at a moment’s notice. In a future directed by Michael Bay, that plays out as a semi-truck reconfiguring itself into a bipedal fighting machine halfway through a front-flip. For now, though, the action is a little more modest. In this clip, we see the real state of the art in self-assembling bots: a bunch of little magnetic cubes scuttling around a tabletop. It’s much cooler than it sounds.
M-Blocks are a new breed of self-assembling robot currently in development at MIT. Each cube is about an inch and a half across on each face, with a flywheel on the inside and an array of magnets on the outside. By spinning the flywheel at high speeds–up to 20,000 revolutions per minute–the self-contained units can scoot across tables and flip themselves through the air. Once they come close to another block, a clever system of self-aligning magnets attaches them to their partner. Seeing a single cube clamber on top of another isn’t especially impressive. But watch several move at once, with disparate parts moving independently and the larger whole rapidly taking a totally new form, and you can start to see a hazy path towards Optimus Prime (A character from the Transformers franchise).
Kyle Gilpin, a researcher who’s working on the project along with John Romanishin and robotics professor Daniela Rus, says there are precedents for various aspects of the design, but the way M-Blocks puts them together is entirely new. There are some bots that use magnets to bond, Gilpin points out, and others that use flywheels for movement, but there aren’t any that use these things towards an autonomously reconfiguring end. Plus, the setup is far more elegant than previous attempts at assembling bots, many of which had unwieldy external components. Gilpin says.
“In general, our system is unique because everything is exceedingly simple. The modules in the video only have two motors–one to spin the flywheel and another to actuate the braking mechanism. Likewise, the bonding mechanism is completely passive. The magnets self-align and naturally draw neighboring M-Blocks together.”
That magnetic bonding system is worth a closer look. Each cube’s face has four magnets, making for a firm attachment when two modules are face-to-face. The edge of each cube has two additional cylindrical magnets, which freely rotate as the cubes approach each other, aligning north and south poles. These edges are also beveled, so when cubes are face-to-face, there’s a gap between these edge magnets; when one starts to flip itself to another face of its neighbor, the edge magnets come in direct contact, forming a strong anchor on which the blocks can flip. The beauty of this passive linking system is that it all happens on the outside of the modules, requiring nothing in the way of electronics or motors to control.
The team is currently working on giving the ‘bots a bit more autonomy. In the video, the cubes are being controlled by hand via a remote control. Gilpin says.
“As a result, it’s difficult to accurately control the flywheel velocity and the exact moment at which the brake is applied.”
If you thought controlling a toy helicopter was tough, try sticking a landing with a lurching magnetic cube. The latest generation of the modules have the computational capacity to automate the movement themselves, and the team is currently hashing out the code to drive it all.
But they’re also thinking towards real world applications, something that will require a more diverse collection of blocks. Some might have a single, more powerful flywheel, allowing them to move several blocks at once, dragging themselves and neighbors as one cohesive unit. The researchers are also considering passive, non-moving blocks battery blocks which could charge neighboring modules, giving the entire system the ability to travel greater distances and climb more challenge obstacles. Lacking motors of their own, these could be tossed around as needed by the other modules, Gilpin says.
That’s just the start, though. The team envisions M-Blocks with cameras or claw-like grippers–specialized blocks that could be carried by others and put in place for other jobs. Romanishin explained in a report from MIT.
“We want hundreds of cubes, scattered randomly across the floor, to be able to identify each other, coalesce, and autonomously transform into a chair, or a ladder, or a desk, on demand.”
When you’ve reached that point, you have to start asking: How many desks does it take to build an Autobot?
COMMENTARY: M-Blocks remind me a lot of Modular Robotics' "Cubelets," toy blocks that snap together with magnets and unlock interesting electronic powers, like sensors, motors, and data displays. I wrote about them in a blog post dated March 8, 2011. The cubelets standard kit comes with 20 magnetic blocks ($300) that can be snapped together to make an endless variety of robots with no programming and no wires. You can build robots that drive around on a tabletop, respond to light, sound, and temperature, and have surprisingly lifelike behavior. But instead of programming that behavior, you snap the cubelets together and watch the behavior emerge like with a flock of birds or a swarm of bees.
M-Blocks are a lot more complicated than the Cubelets because each individual block contains their own logic board, motors, flywheels, breakers and magnets that operate autonomously (on their own without human intervention) allowing the M-Blocks to create patterns of blocks that take the form of shapes like a square, a stack, a pattern, and so forth.
What wasn't too clear in the article and video, is exactly how the M-blocks know what final shape they will take when they go into "construction" mode. The fact that each M-block contains more moving parts than the cubelets described above, raises the issue of durability and mechanical breakdowns. What happens when a fly wheel breaks, or the break motor malfunctions?
M-Blocks are still in early development, and a lot of kinks still need to be worked out. My personal belief is that MIT team should incorporate an industry standard operating system platform so that third party developers can create apps to give the M-Blocks a broad range of functionality similar to the Sifteo Cubes developed by another group of MIT robotic engineers. You can read about the Sifteo Cubes in my blog posts dated March 21, 2011 and August 16, 2011.
Courtesy of an article dated October 9, 2013 appearing in Wired