We’ve seen drones pull off some amazing stunts, like pirouetting through the air while assembling a building brick by brick. But there is something so much more poignant about Cubli, a cube that can stand itself on a single point, perfectly balanced even as the ground beneath it shifts.
Created by researchers Mohanarajah Gajamohan and Raffaello D'Andrea at the Institute for Dynamic Systems and Control, Cubli is a diminutive six-inch cube, stuffed with three coordinated reaction wheels that, through a combination of fluid spins and abrupt stops, can shift the weight and momentum of the block at will. Through these clever mechanics, backed by several sensors and extremely intelligent balancing algorithms, Cubli can jump, balance, and perform a controlled fall.
In practice, this means Cubli can go from standing on its face to an edge, and its edge to a point, balancing itself with all the graceful precision of a prima ballerina. It can also coordinate these three simple movements to roll across a plane, effectively giving the humble cube legs.
But while it may look like nothing more than a cute technical demonstration, Cubli is a tease of a world that’s very closely at hand--one in which objects can not just fly around like R/C toys, but react dynamically to feedback in their environment. When we asked D'Andrea to expand on the importance of devices like Cubli--especially in the coming era of the Internet of Things--he simply pointed us to this presentation he’d given in 2012:
Right now, machines don’t really communicate. If a machine learns something, when you turn it off, it loses that communication. What if a machine could learn and adapt, and share information with other machines? Let me give you an example. If you could create a machine that did a very complex task...and it took a year for it to learn how to do it, you would never deploy it, you would never build it. But what if you had 1,000 of these machines? If you had a thousand of these machines, and they could communicate, and you had the right algorithms, they could learn that task in one day.
As our machines combine the ability to dynamically react to their environment and learn from one another, we will not just have cubes that stand on one corner, drones that play catch, or agile biped robots that beat the DARPA challenge. We’ll have buildings that can build themselves, and objects that can improve themselves. We’ll design things not really knowing what those things could become.
As D'Andrea puts it, such inevitable technology is: “Incredibly powerful...incredibly dangerous…”
[Hat tip: Colossal]