A lot like Voltron, these self-configuring stochastic robots developed at the Cornell Computational Synthesis Lab become a lot more effective when they work together. The bots piece themselves into something larger like a series of robotic Lego blocks as they are moved against one another and, when there's more of them, the combination possibilities become more flexible and varied. That's why the robo-blocks are more effective when they're smaller, as researchers can cram a ton of them in one space.

While most other self-configuring robots will rearrange themselves based on a set plan, Cornell's little boxes use the motion of their environment to come together in ways that can't be precisely predicted, though the bots don't require any local source of power or self-driven movement to operate. This could have some seriously cool applications in medicine and beyond, since the robots could be delivered to where they need to be, en masse, and then build themselves up. After all, a complex centimeter-tall surgery robot would still be less comfortable than a few hundred nano-sized ones, which could just seep into you and form up. — Kevin Hall

BotJunkie, via Cornell Computational Synthesis Lab