MIT's M-block robots are capable of self-assembly

Self-assembling robots are a new breed of a once thought to be impossible machine of another world. Not just hunks of metal and buttons that can pick up and drop objects, more than the robotic carpet cleaner Roomba, and well beyond the mechanic dance move from the 1960s, self-assembling robots have catalyzed a new generation of automatic devices.

The idea was thought to be impossible when a young John Romanishin proposed the design to his robotics professor Daniela Rus in 2011. “I am inspired by a desire to try to design and build interesting systems that could be helpful to society,” Romanishin  said in response to the question of what inspires Romanishin to push scientific limits. She said it would be impossible. However, Romanishin was not convinced. Romanishin and Rus, along with Kyle Gilpin have worked together to defy the impossible.

The execution of this idea was spawned from a young man. In an interview Romanishin was asked how professionals take his ideas and project seriously?  “Nobody really took me seriously until I built a prototype and showed them a video of it working,” Romanishin said. “It also probably helped that this first prototype was also part of an application for a small grant (MIT Eloranta Fellowship)."

"I think that the main lessons that I learned from this, which are now much more clear to me, are that taking initiative really is as important as we are told. In my experience professionals or professors are so busy and are bombarded with ideas and requests for funding, to try to show that you are prepared to attempt to actually implement ideas yourself and attempt to find resources goes a long way.”

These self-assembling robots are small cubes, known as M-blocks, which have no external moving parts. This essentially means that they move unlike animals or humans. No appendages or limbs assist in movement. M-blocks have the unique power to climb over, around, under, roll on the ground, flip through the air, and move while hung from a metallic surface. The mechanism that makes them work is a flywheel that lives inside of the M-block.

Specifically, the flywheel moves at the speed of 20,000 revolutions per minute. When it is put to a stop, the wheel projects an angular momentum to the cube. The edges of the M-Block additionally have magnets arranged on the sides in order for any two M-blocks to connect to one another.

Rus claims that the execution of this idea has been long sought out after in the modular robotics community. She explains that researchers have studied a sliding-cube model where two cubes facing each other can slide up the side of the other and, without changing orientation, slide across its top.

Studying this model helped acquire a sense for the development of self-assembly algorithms. Before, Rus describes how her team had a similar, stable model. This enabled MIT researchers to then manipulate this new robot and significantly simplify the design by leaving the principle of static stability behind. Because of this, the researcher’ robot relies on high-level engineering, enabling faces of any cube to attach to one another.

When asked how Romanishin expects to further the project, he said, “The main interesting research directions that we are interested in are making the modules more capable, (add sensors, add multiple directions of motion) while also decreasing the cost of each module enough to build somewhere from 20-100, and then implement interesting algorithms in an attempt to make this larger number of modules perform useful functions autonomously.”