Purchasing new hardware? Read our latest product comparisons

Self-sculpting smart sand could assemble itself into solid replicas of objects


April 12, 2012

"Smart pebbles" are cubes about 10 millimeters to an edge, with processors and magnets built in  (Photo: M. Scott Brauer)

"Smart pebbles" are cubes about 10 millimeters to an edge, with processors and magnets built in (Photo: M. Scott Brauer)

Image Gallery (3 images)

Research currently underway at MIT’s Distributed Robotic Laboratory (DRL) could lead to an innovative replicative manufacturing technique with the disruptive potential equal to that of 3D printing. Imagine a sand-like material that could autonomously assemble itself into a replica of any object encased within. Incredible though this may sound, the DRL researchers have already managed to build a large scale proof-of-concept, with 10-mm cubes acting as the grains.

Before we go into how these cubes - or "smart pebbles" - work, let’s sketch out the general concept. The idea is to create objects using a subtractive method, where excess material is removed just like when carving in stone. Each grain of smart sand would be a self-contained micro computer. These tiny machines would use an elaborate algorithm to communicate with the neighboring particles in order to establish the exact position and shape of the input object so that it can be replicated.

The already mentioned smart pebbles demonstrate this principle in a more easily understandable 2D setting. First the pebbles establish which of them border on the perimeter of the object to be replicated. Once identified, these particles pass on a message to their neighbors, and effectively specific particles selected by the algorithm are notified that an identical (or scaled) arrangement should be recreated a safe distance away, so that the two shapes do not overlap.

Once the perimeter of the copy is identified, the pebbles within that area bond to each other, while the redundant material simply falls away. The resultant object would be solid, but it could be easily deconstructed simply by putting it back into the heap of smart sand. The constituent grains would detach from each other and the whole process could be repeated with an entirely new shape.

Each smart pebble cube used for testing was equipped with a set of electro-permanent magnets on four sides. The magnetic properties of such magnets can be switched on and off using electrical impulses, but unlike electromagnets, they do not require electricity to sustain these properties over time. With each particle neighboring on eight other particles in a 2D scenario, the magnets allow for selective bonding with any of the neighbors. However, the magnets also play a role in communication and power sharing.

Each smart pebble was also fitted with a rudimentary microprocessor capable of storing 32 kilobytes of code and boasting two kilobytes of working memory. With such limited processing power at the disposal of a single unit, the main computational heft had to fall on the distributed intelligence algorithm that constitutes the core of the current DRL endeavors.

"How do you develop efficient algorithms that do not waste any information at the level of communication and at the level of storage?" asks Daniela Rus, a computer science and engineering professor at MIT. The answer to that question is likely to be found in a paper that Rus co-authored with her student, Kyle Gilpin, and which is going to be presented in May at the IEEE International Conference on Robotics and Automation.

The algorithms developed at DRL have already been shown to work robustly with 3D scenarios, where the bed of smart sand would be divided into layers, each constituting a separate 2D grid. Now the only thing that stops smart sand from joining 3D printing in revolutionizing the world of rapid manufacturing is getting the scale right.

But according to Robert Wood, an associate professor of electrical engineering at Harvard University, this is not an issue. Wood reckons recreating the functionalities of the smart pebbles in smaller scale is feasible. Yes, it would require quite a lot of engineering, but the goal is well defined and reachable. “That’s a well-posed but very difficult set of engineering challenges that they could continue to address in the future.”, he says. If Wood is right, the future of subtractive manufacturing is bright.

Watch the video below to find out more about the algorithm behind smart pebbles.

Source: MIT

About the Author
Jan Belezina Formerly in charge of Engadget Poland, Jan Belezina's long time fascination with the advance of new technology has led him to become Gizmag's eyes and ears in Eastern Europe. All articles by Jan Belezina

2048 bytes? What century is this article from? That cube has way more than 10 times the volumetric space that my 137,438,953,472 byte (128gig) USB thumb drive chip uses.

There's 2 ways to build a CPU - with loads of custom circuitry and instruction sets etc, or, with an almost completely brain-dead minimist circuit that draws on memory. Seems like they chose the wrong one...


You haven't got what is all about...

It's not about how you can minimize digital storage media or build a processor. Those tiny elements don't need high computing power and the whole idea is about to make self-organizing mobile structure that able to autonomously copy a shape. Can your processor or USB chip do that, wise guy?

Iván Imhof

This looks to be focusing more on the algorithm than the physical side of things; but I wonder what the tensile strength of this assembled "material" would be given current technologies?


I like this conecpt but I might be missing the point here. I understand how they are able to replicate a shape (seems like a straightforward algo).

I like the video where the connections are broken between the non-relevant parts those creating a replica. question is how are those connections created in the first place, and what happens with the 'waste'. Can it be reconnected for future use?

David Codish

They are working at a 'proof of concept' level of research and development. There's not going to be the greatest efficiency of design at that level as it's more important to be able to easily modify the design and test than to optimize the design. Once there is proof the concept works, funding for development of a prototype with actual utility can sought. There is a reason for the delay between conceptual development and actual product useable in the 'real world'.


I wonder what each of these grain of sands would cost. Then again this would probably not be used for static objects, so cost would maybe be divided over all built objects.

And I'm guessing the strength of these constructions will be a limiting factor for it's potential.

Just thinking out loud ;)

Patrik Nordberg

If they used this technology to recreate the bordering area then theoretically you could create a purposely shaped environment in which to grow an object... Ie: An engine could be immersed in this 'sand,' (section by section mind you not a complete motor) it's exact border shape replicated then from the resulting 'mould' that was created, a replica formed. Translate this to medical applications and designer organs / bones / muscles could potentially be moulded based on the 'original.' Identity theft could take on a whole new level!

Andrew Donaldson

Likely a first step toward Kurzweil's Utility Fog! (a.k.a. my favorite of his predictions)



Nathan Brothers
Post a Comment

Login with your Gizmag account:

Related Articles
Looking for something? Search our articles