By Doug Gross, CNN
It could be something out of "Harry Potter," or a scene from "Terminator 2" if you want to take it to a creepier place.
Take a box full of sand and tell it what you need - say a hammer, a ladder or a replacement for a busted car part. Bury a tiny model of what you need in the sand, give it a few seconds and - voila! - the grains of sand have assembled themselves into a full-size version of the model.
MIT robotics researchers say such a magical sandbox could be no more than a decade away.
A team from the school's Computer Science and Artificial Intelligence Laboratory says they've developed algorithms that could enable "smart sand" - essentially miniscule, simple robots that would communicate with each other about how to align together properly once they've been given a model to copy.
The team has already done limited testing with larger cubes - 10 millimeters wide with rudimentary microprocessors inside and magnets on four of their sides. The "robot pebbles" magnets are used not just to connect, but to communicate with each other and share power.
"The 'robot pebbles' are not going to turn into true 'smart sand' overnight - but it will happen ...," said Kyle Gilpin, a graduate student working on the project.
Gilpin, who authored the paper the team will present at next month's IEEE International Conference on Robotics and Automation, predicted it could take 10 years, but that "we'll see incremental improvements along the way."
The grains of "sand" would essentially work together like the block of stone a sculptor begins with. Once deployed, the grains needed to build an item would move into place, while those that aren't needed would simply fall away.
“Say the tire rod in your car has sheared,” Gilpin said. “You could duct tape it back together, put it into your system and get a new one.”
Once an item is no longer needed, the grains could be ordered to fall apart and get ready for the next project.
One of the main challenges at this point is getting enough computing power onto items so small. The "robot pebbles" now being tested each have a tiny microprocessor that can store just 32 kilobytes of program code and have two kilobytes of working memory.
But Gilpin said that's not a reason to lose hope.
"Consider how rapidly and dramatically computers have been miniaturized over the last 50 years," he said. "What used to
occupy an entire room now fits on a small fraction of a fingernail. We'll see the same advances applied to programmable matter systems as well."