Home / Science / Scientists have just found a way to allow machines to see through clouds and fog

Scientists have just found a way to allow machines to see through clouds and fog



Whether it’s robots working in a disaster area, autonomous cars moving around cities, or satellites peering into space, having machines that can see through clouds, mist and fog is incredibly useful and scientists may have just created the best system so far.

The newly developed system works via an algorithm that measures the movement of individual light particles or photons, as emitted in fast pulses by a laser, and uses them to reconstruct objects that are obscured or hidden from the human eye.

What makes the technique even more special is the way it can reconstruct the light that has been scattered and bounced off the barrier onto the road.

In the experiments, the laser sight was able to see objects hidden behind a 1

-inch layer of foam.

laser scanning(Stanford Computational Imaging Lab)

“Many imaging techniques make the images a little better, a little less noisy, but this is really something where we make the invisible visible,” says electrical engineer Gordon Wetzstein, of Stanford University.

“This is really pushing the frontier of what could be possible with any kind of detection system. It’s like a superhuman vision.”

As the laser light passes through the barrier – the foam, in this study – only a few photons hit the object behind, and even fewer do it back again. However, the algorithm is smart enough to use those little bits of information to reconstruct the hidden object.

Officially, it’s known as confocal diffuse tomography, and while it’s not the first method for looking through barriers like this, it offers several improvements: it can work without knowing how far away the hidden object is, for example.

The system is also capable of operating without relying on ballistic photons, as other approaches do: these are photons that are able to travel to and from the hidden object through a scattering field, but without being distorted.

“We were interested in being able to create images through scattering media without these assumptions and to collect all the photons that were scattered to reconstruct the image,” says electrical engineer David Lindell, of Stanford University.

“This makes our system particularly useful for large-scale applications, where there would be very few ballistic photons.”

Large-scale applications such as driving a self-driving car in pouring rain, for example, or even imaging the surface of the Earth (or other planets) through cloud haze – there are many potential uses here. Researchers are eager to continue experimenting with more scenarios and more scattered environments.

Current systems are not particularly good at handling light scattering caused by fog and haze.

LiDAR, for example, is brilliant at detecting objects the human eye can’t see, but it starts to have problems when rain or fog interfere with its detailed laser scans. Later, this system may solve the problem.

Before we get ahead of ourselves, it’s worth noting that scans using this method can take anywhere from a minute to an hour, so there’s still a lot of optimization to work on.

That said, recreating a hidden object in three dimensions that the human eye cannot see is an extremely impressive feat.

“We are excited to go further with other types of scattering geometry,” says Lindell.

“So, not just objects hidden behind a thick sheet of material, but objects that are embedded in densely scattered material, which would be like seeing an object surrounded by fog.”

The research was published in Nature Communications.


Source link