Researchers have created magnetic, robotic blocks that can pivot, jump, and lock together to create various larger, collective shapes. The blocks are fitted with on-board power, computation, actuation, and communication capabilities. A flywheel and braking mechanism allow each block to move independently from one another. Kyle Gilpin, 2013
Researchers at the Massachusetts Institute of Technology have created magnetic, robotic blocks that can spin, jump, and lock together to form larger shapes—with no external moving parts.
October 15, 2013—Imagine a world in which swarms of miniature robotic blocks can travel over difficult or hazardous terrain, and upon reaching their destination lock together to form any necessary shape to solve the problem that they have been dispatched to resolve. That future is now one step closer. Researchers at the Massachusetts Institute of Technology’s Computer Science and Artificial Intelligence Laboratory have created prototypes of magnetic, robotic blocks that can spin, jump, and lock together to form larger shapes—all with no external moving parts.
The magnetic, robotic blocks, dubbed M-blocks, measure 50 mm on each side and weigh 143 g. (See a video of the blocks.) The framework of each block has been milled from a single piece of aluminum, according to John Romanishin, a robotics engineer at the lab who wrote in response to questions posed by Civil Engineering online. At the edges that lead to each corner, the cubes hold cylindrical bonding magnets that have been placed so that they can spin freely, similar to how a rolling pin operates. These 24 magnets can thus reverse their polarity as necessary so that any block can lock together with any other block.
Each of the cube’s six bolted-on faceplates contains eight smaller magnets to help align the blocks as they lock together to form their larger collective shapes, but the cubes also boast beveled edges that allow the edge magnets to lock together on a diagonal, creating a bond that is stronger than a face-to-face bond.
The momentum that breaks the magnetic bond that links the cubes together and propels the block forward is created by a flywheel encased within each cube that rotates at a rate of up to 20,000 revolutions per minute before a braking mechanism engages. The torque caused by the braking allows the blocks to walk, pivot, and jump together to create various larger shapes. (See a video of the process.)
“I can't really remember exactly where the idea came from, but three of the important concepts have been with me for a while,” Romanishin said. “Like many engineers, I obsessively played with Legos as a kid, and I have always thought that more or less everything in the world should be modular and reconfigurable and work more like Legos,” he explained. “I am also somewhat obsessed with magnets, and it is a natural progression to try to combine modular devices with magnets.” The final piece of the puzzle came from a freshman physics class Romanishin took at MIT in which the class was taught how gyroscopes work by rotating a spinning bicycle wheel while sitting on a swiveling chair. “I was just fascinated that powerful forces could seemingly come out of nowhere due to angular momentum interactions,” he said.
Several years ago, the three ideas coalesced to form the idea for M-blocks.
Each of the prototypes measures 50 mm on a side, weighs
143 gm, and has been milled from a single piece of aluminum.
Six bolted-on faceplates, each of which contains eight small
magnets, enable the blocks to self-align, even when cantilevering.
John Romanishin, 2013
Moving from the theory to reality took a collaborative team, however. While Romanishin has worked on the mechanical engineering of the blocks, Kyle Gilpin, Ph.D., an electrical engineering post-doctoral student in the laboratory, has been responsible for the robots’ electronics and algorithmic design. Daniela Rus, Ph.D., a professor in the electrical engineering and computer science department at the university and the director of the lab, is overseeing the work.
Currently, the prototypes can be moved with a remote control joystick or by preprogramming their computer chips with a series of commands. “There is some intelligence in the cubes, but there is not complete autonomy at this point,” Gilpin says. “But that is a direction we want to go in—we’d essentially like to be able to toss a bunch of cubes out on a floor, have them locate one another, move toward one another, coalesce, and then form some particular shape.” Currently the single flywheel in the cubes allows them to move in one direction, but the team is working on next-generation prototypes that will enable them to move in any direction.
Practical applications in the future could include having swarms of blocks temporarily repair infrastructure in inhospitable or difficult-to-reach locations, or perform reconnaissance missions, looking for survivors in disaster zones, Gilpin says. The simplicity of the cubes’ design—no external moving parts and the ability to jump and change their collective configuration—means that they can travel over very rough terrain or through rubble without the difficulty that a more traditional robotic system would experience, Gilpin notes. A camera embedded in one or more of the cubes could transmit images back to the operator.
“The cubes themselves are robust and they’re great in compression,” Gilpin notes. “If you have a whole bunch of them together, the shear strength is obviously not as strong as a solid piece of material, but we’re actually impressed with the individual unit strength.” Although additional research is needed, the compressive and shear strength of future swarms of cubes could be manipulated by changing the type of material used to mill the cubes’ framework and the method of bonding the cubes together, according to Romanishin.
Eight prototypes currently exist, according to Romanishin: four active and four inactive. The active modules are fitted with on-board power, computation, actuation, and communication capabilities, while the inactive modules are used as the base cubes from which larger shapes are built. These have not been fitted with robotic interiors.
Currently the prototypes each cost approximately $300 to produce, a cost driven primarily by the amount of milling work that is necessary. “However there is nothing fundamentally all that expensive about any of the components, and we hope to make our next version both much more capable and, hopefully, cheaper—in the $50 to $100 range,” Romanishin said. Mass-producing the cubes in quantities of 100,000 or more would bring the cost down even further, he said.