Deep learning enables real-time imaging around corners:
Researchers have harnessed the power of a type of artificial intelligence known as deep learning to create a new laser-based system that can image around corners in real time. With further development, the system might let self-driving cars "look" around parked cars or busy intersections to see hazards or pedestrians. It could also be installed on satellites and spacecraft for tasks such as capturing images inside a cave on an asteroid.
"Compared to other approaches, our non-line-of-sight imaging system provides uniquely high resolutions and imaging speeds," said research team leader Christopher A. Metzler from Stanford University and Rice University. "These attributes enable applications that wouldn't otherwise be possible, such as reading the license plate of a hidden car as it is driving or reading a badge worn by someone walking on the other side of a corner."
[...] The new imaging system uses a commercially available camera sensor and a powerful, but otherwise standard, laser source that is similar to the one found in a laser pointer. The laser beam bounces off a visible wall onto the hidden object and then back onto the wall, creating an interference pattern known as a speckle pattern that encodes the shape of the hidden object.
Reconstructing the hidden object from the speckle pattern requires solving a challenging computational problem. Short exposure times are necessary for real-time imaging but produce too much noise for existing algorithms to work. To solve this problem, the researchers turned to deep learning.
"Compared to other approaches for non-line-of-sight imaging, our deep learning algorithm is far more robust to noise and thus can operate with much shorter exposure times," said co-author Prasanna Rangarajan from Southern Methodist University. "By accurately characterizing the noise, we were able to synthesize data to train the algorithm to solve the reconstruction problem using deep learning without having to capture costly experimental training data."
More information: Chris Metzler et al, Deep-Inverse Correlography: Towards Real-Time High-Resolution Non-Line-of-Sight Imaging, Optica (2019). DOI: 10.1364/OPTICA.374026
(Score: 2, Insightful) by Anonymous Coward on Saturday January 18 2020, @12:21PM (1 child)
It seems to me that bending light around a corner with what is effectively same as using mirrors. Is still a subset of line of sight.
(Score: 0) by Anonymous Coward on Saturday January 18 2020, @02:55PM
Adding a mirror isn't adding line of sight. Line of sight means unimpeded and unreflected. Bouncing off a mirror means there's two lines of sight.
(Score: 1, Interesting) by Anonymous Coward on Saturday January 18 2020, @12:32PM (2 children)
How does that work, or is the paint coating big enough to read license plates with this method? Sounds pretty bullshit
(Score: 2) by SomeGuy on Saturday January 18 2020, @01:56PM (1 child)
ANYTHING to do with AI is bullshit. Now they are just literally using smoke and mirrors.
(Score: 0) by Anonymous Coward on Saturday January 18 2020, @02:22PM
I'm pretty sure they can get a generic view of item existing behind the curve, but the:
such as reading a car plate
part is pure retarded bullshit
(Score: 2) by hendrikboom on Saturday January 18 2020, @01:08PM (1 child)
I wonder if this would still work if the wall were lumpy, or in a different place from where it was during training, or if the wall were a moving, shape-changing person instead of a stationary wall.
(Score: 0) by Anonymous Coward on Saturday January 18 2020, @02:51PM
The basic process works fine with irregular reflectors (and in fact needs some at the fine scale, to not be a NOP - else it'd be a plain mirror without a speckle inversion step).
The reflector can't be something quickly moving, like a person, with the setup they demonstrate, which needs a 1/8th second exposure. But that's not inherent to the optics nor math, just to their sensor and computation.
(Score: 4, Interesting) by Hyperturtle on Saturday January 18 2020, @03:06PM (3 children)
What is special about being zapped repeatedly with a laser that won't make pedestrians go blind when a reflected laser hits everyone repeatedly all the time in an urban environment?
These things rely on using reflections; that means the surfaces being used to provide the reflection will be variably uncontrollable and the laser light can't be assumed to be at any given height; lasers may end up bouncing off airplanes. Maybe the power source is too low to go very far or cause much damage, but those details aren't mentioned.
The article and the "more information" both do not contain reassurances about the laser type used; I concede that the "laser pointer" reference may have been chosen as a term that a casual reader would understand as a comparative concept. As a casual reader, I know people go blind or go to jail when shining those laser pointers accidentally or inapprorpriately.
Considering that it's a DARPA funded project, I have to think that collateral damage might even be a positive as part of passive defense system on any equipped vehicle. Non-lethal harm is always a positive when the collateral damage is a public relations issue. Besides, I imagine that it'd take repeated exposures, so field testing with DARPA out in some mocked up environment will not really capture data comparative to a densely populated civilian urban environment filled with self-driving cars (that become more prevalent when people start having vision problems and rely on cars with this feature to get around... my apologies for the gallows humor.)
I don't have any anything to suggest my concerns are valid, but maybe it's all hype and spin and that's why the focus is on using a computer with lasers to do it, because that is a good way to secure interest and funding. I'm surprised there weren't any clouds mentioned--maybe fog makes it hard for the system to see? Lots of reflections on wet surfaces, mists, and dust... makes me wonder how it'd work in a dusty desert or a rainy cityscape.
(Score: 2) by janrinok on Saturday January 18 2020, @05:35PM
It depends on the frequency of the laser. Invisible (to man) lasers already exist, they are not all bright green.
(Score: 2) by darkfeline on Saturday January 18 2020, @10:32PM (1 child)
The same way pedestrians get zapped by the reflections of the light from the sun without going blind? You do realize that lasers don't have to be "burn a hole through your face" high power right?
Join the SDF Public Access UNIX System today!
(Score: 1, Touché) by Anonymous Coward on Sunday January 19 2020, @08:50AM
If they are not "burn a hole through your face" then what is the point?
(Score: 5, Insightful) by jmichaelhudsondotnet on Saturday January 18 2020, @06:58PM
we are being spied on more than we can even possibly know and it is just gettting worse.