CES-goer says his camera was killed by a self-driving car's LIDAR


Originally published at: https://boingboing.net/2019/01/20/1550nm-considered-harmful.html





Gotta say, future, melee of robo-cars with laser weapons sounded much cooler in the brochure.


That’ll learn ya.


FFS, even the most basic fact checking dispells this story. All modern digital cameras have IR block filters. Silicon–the material used in the sensors for everything since the vidcon era–is transparent to IR. If a camera lacks an IR filter, then all the sensors will respond to IR emissions which show up as white (or purple) artifacts in the image. Helful hint, the sun products a lot of IR.

So, to assert that his camera was damaged by an IR laser in the low power range that is allowed for use in LIDAR and similar systems is just absurd.

There are special digital cameras that lack the IR block filters, but they’re designed for specific scientific uses–where you really know what you’re doing and want to record the IR content of a scene. It’s possible he had such a camera, but, the last time I looked they were $5K+ as they were special devices made for people with deep pockets. Let’s just say he had such a device. Well, idiot, you did something stupid.

Any way you look at it, this story doesn’t pass the sniff test. Cory should know better than to repost this crap. Should…


I’m a little surprised that a visible-band camera sensor extends so far into the IR band, up to the eye safe wavelengths. Also optical glass often is opaque to IR. Probably a chance combination of the close range, intra-beam viewing and the magnification of a comparatively weak signal by the high-end camera lens. main-qimg-db85078977ec318ee7bc39c5ca0d84b4


It’s kind of astonishing that it’s the pedestrian Class 2 “red dot” laser in a (properly operating) laser cutter that’s the dangerous one. The powerful Class 4 laser that can cut through wood and plastic is completely neutralized by a humble pane of glass.


Indeed, but anually the Class 2 laser results in more injuries bscause people frequently choose to purposefully defeat their blink reaction (“I bet you cant stare at this for 10 seconds, I double dare ya!”). A Class 4 laser has to be protected with interlock mechanisms to prevent operation if someone were to remove that humble pane of glass.


Just because the camera has IR filters doesn’t mean it still isn’t more sensitive to IR than human eyes. You can prove it easily - look at the front of a TV remote control while mashing its buttons. Then do the same through a camera’s viewfinder.


meaning don’t look at the sun even if it is flatish

1 Like

and because this can NEVER happen, Aeye

has offered to replace it (the damaged camera)


more in-depth:


btw; there tons of videos out there, which shows damaging sensors in concerts by eye-protective-show-lasers while filming and those are even in the visible spectrum:


as the photographer put it:

“This sensor damage was an effect of a combination of things — intensity, amount of time, spot size, wavelength, pulsing"


I’m not sure why you are so confident of the IR filters.

Yes, they definitely have them because the color balance would be a disaster otherwise; but it’s common for even feeble remote control IR and similar to still be visible. Against a nice fiber laser, advertised as being suitable for lidar at up to 1km, while standing right next to it?

Seems at least plausible that you might get a thermal kill on a few pixels, though certainly less efficiently than if it were a deliberately IR camera.


Not true for near IR, in the 1,000-1,500nm range. These wavelengths are especially dangerous to human (and animal) vision, as the light ficuses on the retina, and being invisible there is no blink or aversion reflex. You don’t know to look away.

For far IR, in the 10um range (10,000nm), most glass and plastic will efffectively block the light.


It sounds like AEye is dancing with the safety rules. “Totally can’t blind humans, unless they’re wearing glasses, or have some sensitive condition, had cataract surgery or…”

If AEye claims that their system follows the safety rules and yet still burns out camera sensors, what happens when an equiv-tech system burns out their sensors?


Considering I hadn’t read the photographer’s write up (shame on me) it sounds like I was pretty spot-on.

I once damaged a laser power meter when I accidentally introduced an errant refraction through a piece of plastic placed intra-beam. The meter was rated for the power and wavelength but the accidental lensing from the plastic made an intense beam sufficient to cause a minor burn. So random, I couldn’t even repeat the results when I tried.


Without going into a lengthy lecture, I’m just going to say that you need to learn a lot more about light and optics before making such claims. A couple of bullet points:

  • A laser powerful enough to to be used for LIDAR at 1km distances is dangerous to living eyes and can most certainly damage video sensors.
  • IR filters in cameras (if they are actually present - please cite some evidence) would not be optical density of 5+ or so, which is what would be needed for a laser of this power range.
  • The sun does emit IR, but it is not a virtual point source. Kind of irrelevant, as the visible light from the sun coming though a camera lens (or an eye) is sufficent to cause damage with or without the IR.

The article is a good warning about laser LIDAR being used in vehicles. Yes, it can damage cameras, but we should be far more concerned about the potential damage to our eyes.

For what it’s worth, I have a degree in laser technology and have been working with industrial lasers for 30+ years.


We should also worry that different animals have eyes that vary a lot in sensitivity. A car safe for humans may still do lots of damage to insect or whatever.


Might be good technology to roll out if LIDAR like this can also render red light cameras inoperable.

1 Like

Seems unlikely. In that case the LIDAR points away from the camera because they both focus on the same object.