One day, the researchers imagine that you will be able to deposit an object into a box of sand-grain-sized computers and pull out a full-size replica of the original object a few seconds later. (3D printing, eat your heart out!)

The researchers are currently experimenting with a larger-scale version of their sand-computers with 10-millimeter cubes. Each cube is equipped with a rudimentary microprocessor that can run 32KB of code with just 2KB of working memory.

The cubes start off in a large block formation that's held together with magnets on four sides of each cube--two sides were left out to make room for the microprocessor and circuitry to regulate power usage. The magnets themselves are "electropermanent" in nature, meaning they only need a single electric pulse to switch between magnetized or demagnetized. While the cubes are connected, they can communicate with each other and even share power.

When the block of cubes needs to reform itself into another object, it disengages unnecessary parts, which then simply fall away from the rest of the mass. The building method uses a subtractive process, which is the opposite of the usual additive approach used by most reconfigurable robots and 3D-printers.

Defining the object you want to duplicate is a little more complicated since you first need to bury it inside of the pile of smart cubes. While the object is inside of the block, the cubes communicate with each other to determine the perimeter of the object. From there, it duplicates the object's shape in another section of the block, while the rest of the mass disengages itself, leaving you with only the replica and original. (Skynet, anyone?)