For a robot to solve a Rubik’s cube, it needs to be able to do three things: look at the cube and understand what it’s seeing, determine the rotations it needs to perform to solve it, and actually manipulate the cube. As humans, we consider the first and last tasks the easy ones. It’s simple enough to identify the colors and spin the sides, but really difficult to work out the steps to get to a solution. For robots, finding the solution is as straightforward as following an algorithm, but processing the visuals and handling the cube is tricky.
That’s what makes this robot so interesting. It uses a high-speed camera and computer vision processing to quickly determine the layout of the cube, and then just three fingers on its “hand” to rotate the necessary side. It uses two of those fingers to flip the cube so the side it needs to rotate is up, and then firmly holds the cube in place. Then, it uses the third finger to spin that side. It repeats the process over and over, until the Rubik’s cube has been solved a couple seconds later.