Cute yet creepy? Somersaulting MIT cube robots can self-assemble

These brightly colored blocks may look like child’s toys, but watch out: These adorable cubes are actually spinning robots that can connect to build modular machines. There’s no assembly required: The blocks, built at MIT, will do it themselves.

The acrobatic boxes, called M-blocks, have no external parts. And yet they can spin, somersault and snap together to create all kinds of shapes, depending on the job at hand.

“It’s simple on the outside, but the insides are very unique,” MIT robotics professor Daniela Rus said in an interview.

Unlike a normal robot that can’t change its form, these modifiable robots can be made into any shape ideal for a given task’s needs, reconfiguring at a moment’s notice, Rus said.


It’s almost like a cuddlier, more basic version of the T-1000 — the shape-shifting, liquid-metal nemesis in “Terminator 2.”

“I just had the idea for a really long time,” said MIT roboticist John Romanishin, who proposed the idea while a senior at the university. “The idea has been in a lot of people’s minds from movies like ‘Terminator’ and other popular culture references.”

The blocks move thanks to a flywheel inside their bodies that can spin at a blistering 20,000 revolutions per minute and then can brake straight down to zero in a blinding 10 milliseconds, said MIT roboticist Kyle Gilpin. When the flywheel suddenly stops, all the energy from that angular momentum transfers into the cube’s frame, causing it to flip.

“A low amount of energy will cause it just to roll forward, an intermediate amount of energy might cause it to climb a wall and the highest amount of energy will cause it to do something like jump,” Gilpin said.

The boxes have magnets on the edges that can grip their cube-mates, allowing one to flip around into a new configuration without breaking the connection.

The cubes are made of 2-inch blocks of aluminum hollowed out to make the boxy frame, filled with the motor and brains, and then plated with colorful plastic sides 3-D-printed at the lab. They’re very sturdy, Gilpin said.

“John likes to abuse the cubes,” Gilpin joked. “He likes to stand on them and drop them.”

The bright tins on the boxes, while pretty, even serve a function, Rus said.

“The colors are useful for debugging because you can see what each one does when you [test] it,” she said.

Ultimately, the goal would be to make smaller, millimeter-sized versions of these robots, Rus said. And the bots should have a number of applications, from building temporary bridges after a collapse to probing a suspect pipe.

“A single cube can use this tumbling motion to traverse very hard-to-reach terrain, so maybe tunnels or through rubble,” Rus said. “So it could jump around in these kinds of difficult environments.”

The researchers are currently sending the cubes commands via a radio signal, but they hope to put the algorithms directly into the cubes so they can decide themselves how and where to move.

And no, doing so won’t turn “Terminator 2" into reality, the researchers added.

“It’s a very interesting and capable robot, and the more it can do by itself, the less human involvement is needed,” Rus said. “We think these cubes can really be focused on applications that would actually help.”


Eureka! Did crashing neutron stars forge all gold in Universe?

A hunt for dark matter in a former gold mine

50 shades of window? ‘Smart glass’ blocks heat, light at flip of switch