Photos and videos are not lightweight files — they quickly add up to gigabytes of data which can be a dealbreaker a lot of research. Engineers at the Swiss company iniLabs created a better way — a camera that borrows its mechanics from the marvels of the human retina.
The Dynamic Vision Sensor (DVS) works a lot like the human retina which makes for a hyper efficient and ultrafast camera. The individual neurons in our eyeballs don’t actually record all of the information in our field of view; they just spot the changes in movement. This gets rid of lots of extraneous data from the surrounding scene. It’s also exactly how the DVS works. By selectively recording only the motion, the DVS can record hours of video using very little power and only a few megabytes of data storage.
The biological connections goes beyond mimicking the retina, though. The chips that power the DVS are actually from IBM’s line of brain-inspired chips, the TrueNorth computer architecture. It only makes sense to support one human-like process with another, say the researchers that developed the system. “What we’re talking about — the cameras sending information when something changes — is actually a very central theme to how the brain works, or at least how neuroscientists think it works,” Cornell computer scientist Nabil Imam told the Technology Review. “We’re capturing brain features at a high level.”
But frankly, the DVS works better than the brain sometimes. Because it spots changes in movement almost instantaneously, the camera can provide the visual data needed for systems that require quick reaction times, like this robotic goalie. What it gains in speed and efficiency, though, the DVS loses in resolution. As you can see from this video of milk drops, it takes a pretty grainy picture. [Technology Review]
