For a drone to navigate autonomously requires it to have an extremely good sense of its own location. That’s especially tricky indoors, where GPS — the technology used by all commercial drones — doesn’t work. One way around this is to use onboard sensors, such as lasers or cameras. However, lasers are expensive and consume a lot of power, while cameras also have limitations in low-light situations. In such low-light scenarios, drone motion has to be slowed or else the image will blur.
So what’s the answer? According to researchers at the University of Zurich, the solution lies in their specially developed high tech “event camera,” which works very differently to regular cameras.
“Instead of capturing the light of the scene, an event camera only records changes in brightness for each pixel, ensuring perfectly sharp vision even during fast motion or in low-light environments,” Professor Davide Scaramuzza, director of the University of Zurich’s Robotics and Perception Group, told Digital Trends. “This means that if you point an event camera at a scene that is not moving, it won’t show you anything at all. But as soon as the camera detects motion — for example, if you move the camera — it will show you just that motion on a per-pixel basis and at a very high refresh rate.”
Because the camera looks for pixel changes, it is very sensitive in low-light settings, and also avoids getting confused in extreme bright light. In short, it means that a drone equipped with this camera sensor can navigate in extreme light conditions where others would fail. “Our tests so far have been very successful,” Scaramuzza continued. “We have demonstrated that we can continue to fly even after the light has been completely switched off in a room. In the same situation, a drone using a conventional camera would crash.”
At present, Scaramuzza said the research lab is speaking with drone companies about the possibility of commercializing the technology. To reach this point will mean additional testing — particularly outdoors in places where the drone will have to deal with rapid transitions between bright sunlight and shadow, which are also challenging for regular camera-equipped drones to negotiate.
A paper describing the work has been submitted to the journal IEEE Robotics and Automation Letters.
