How does a robot “see” the world? The obvious answer is that they see it however we humans tell them to. We feed them the video footage that serve as their “eyes,” and then we go on our merry way. But a new video called Robot Readable World makes the argument for a slightly more complex form of robot vision.
Robots and computers currently process the video that we give them using complicated algorithms, and that software is starting to give computers and robots their own distinct way of seeing. We still use cameras to serve as robot eyes, but the bots can now use their own methods to make sense of the footage we provide them. And those methods are, frankly, a little bit creepy.
These kinds of algorithms are already in our lives; they’re the same type that the Microsoft Kinect uses to track your movements, for instance. But much more powerful forms of this software let computers recognize faces, cars, people, and street signs inside video footage.
In one example from the video, computers see auto traffic patterns as procedurally generated arrows that show the direction of moving vehicles. In another example, the computer sees faces as a series of constantly moving ovals. Timo Arnall, the video’s creator, calls it “an experiment in found machine-vision footage, exploring the aesthetics of the robot eye.”
Robot Readable World compiles over 30 different examples of video recognition technologies (all listed on the video’s description), and puts them over an unnerving soundtrack (a song called “Cold Summer Landscape” by the band Blear Moon) to create a vision of how computers break complex forms and movement down into something they can understand.
The interesting thing about these technologies isn’t how much they let computers see things the same way we do; instead, it’s how differently robots see the world. Seeing the kind of intense processing that it takes for a robot to do a simple task like turning a corner really drives home how, even as robots gain the ability to replicate more and more human actions, the machines we’re building are “thinking” in fundamentally different ways.