3 ms·
It appears that only cameras that take close up images of the sensor are affected (from the article). Traffic cams and dash cams are too far away to be affected
by goku12 11mo ago
It appears that only cameras that take close up images of the sensor are affected (from the article). Traffic cams and dash cams are too far away to be affected (you can see this in the posted video itself). The self-driving cameras are wide-angle cameras and are not affected. Therefore, they don't really affect other people's property.
And in case you really need to image it up close, the laser operates at NIR (1550 nm). Our eyes are insensitive above 700nm (which is why they don't damage our eyes), so there isn't any point in capturing that wavelength. You can safely filter it out with appropriate hardware filters, without any sort of loss in image quality.
Now there is a question as to why the phone camera captures a wavelength that we don't see anyway. You can see this even with a regular TV remote (or anything similar with an IR emitter). You can't see the signal directly, but they turn up in captured videos with a purple hue. I haven't explored the technical reason for this. But my guess would be that the camera's internal red and blue filters have unwanted secondary transmission peaks at those wavelengths. So adding an extra wideband blocking filter at that wavelength to every camera by default is not a bad idea.
Another thing about the camera that surprises me is that the sensor pixels can be saturated enough to damage it permanently. I think that these pixels are suffering a permanent stuck-at-1 failure here. Sensor saturation is a common phenomenon and results in washed out images. But this is the first time I'm seeing a permanent breakdown. (Also indicates how much I know (or not) about cameras.)
And of course, the biggest question is what can be done with the Lidar optical power output. That is decided by the required signal-to-noise ratio of reflected light at the Lidar input, for an object at the maximum desired distance. There isn't much you can do here since that range is necessary for safety (so that dangers can be detected at far enough distances). Another parameter that can be changed is the laser wavelength. I don't know much about this either, but my guess is that it's limited by the solid state laser technology. I don't know how much freedom is available on that side.
- MarkusQ 11mo agoThe dash cam, backup camera, or autonomy system of an adjacent car stopped in traffic would be roughly the same distance from the lidar as a person taking a closeup picture. Since these are lasers the attenuation with distance will be far less than with, say, a headlight (that's why they use them). You can't just assume, as the article does, that the inverse square law will protect you. In any case, the problem is acting as if it is the camera owner's responsibility to avoid damage. How would you feel if they mounted a high pressure water cannon on each car and advised people to keep their distance or wear a raincoat if they didn't want to get wet?
- stupidFugger 11mo agoWhat?! Just because you cant see it, doesn't mean it isnt burning your eyes! IR lasers can be incredibly dangerous for this reason.
- goku12 11mo agoThey're saying that the laser at that wavelength is absorbed by the liquid in the eyeball (either aqueous humour or viterous humour). It doesn't reach the retina to burn your eye.
- stupidFugger 11mo agoCorrect! It is absorbed by your cornea, lens, iris. It burns those things instead. No big deal. Because your eyes are insensitive to the light, your iris does not contract meaning more light gets in that would otherwise. Source: Grade 12 physics A textbook I have on lasers Any stryopryo video
- goku12 11mo agoNeglecting what someone says and flaunting the 12th grade textbook instead isn't the flex you think it is.
- faidit 10mo agoChronic IR exposure is a known occupational hazard that can cause cataracts. https://en.wikipedia.org/wiki/Glassblower's_cataract https://en.wikipedia.org/wiki/Glassblower's_cataract In the near future, you might not be able to look at a busy street without making visual contact with dozens of lidar scanners. It's already hard to avoid them in SF. The video referenced in TFA was filmed from about as close as you might be to a scanner while walking on the sidewalk beside a lidar-equipped vehicle. Cars can get very close to pedestrians in busy cities, e.g. at pedestrian crossings. Exposure is much stronger at close distances. Therefore I think there is a significant risk here that has not been studied or tested for in depth.