Where I live I am often surrounded by Waymo vehicles... is Lidar 100% safe for people to be around? I ask because I read an article about how Lidar on one of the new Volvos could destroy your phone camera if you pointed it at it? If Lidar can do that to a phone camera, can it hurt your eyes?
Your eyes will be fine (assuming that we are talking about automotive LiDAR specifically).
Automotive LiDAR is designed to meet Class-1 laser eye-safety standard, which means "safe under normal conditions." It isn't some subjective/marketing thing, it is an official laser safety classification that is very regulated.
However, if you try to break that "normal conditions" rule by pressing your eyeball directly against an automotive LiDAR sensor for a very long period of time while it is blasting, you might cause yourself some damage.
The reason for why your phone camera would get damaged, but not your eyes, is due to the nature of how camera lenses work. They are designed to gather as much light as possible from a direction and focus it onto a flat, tiny sensor. The same LiDAR beam that is spread out for a large retina can become hyper-concentrated onto a handful of pixels through the camera optics.
Why wouldn’t your eye lens focus LIDAR photons from the same source onto a small region of your retina in the same way that a phone camera lens focuses same-origin photos to a few pixels?
Sorry if this is a silly question, I honestly don’t have the greatest understanding of EM.
It's incredibly important to understand that eyes and glass have different optical properties at these wavelengths. It's hard to conceptualize because to us clear is clear, but that's only at visible light. The same way that x-rays and infrared and other spectra can show things human eyes can't see, or can't see things visible light can see, it's a 2 dimensional problem. The medium and the wavelength are both at play. So, when you have the eye which is known to absorb such light, and artificial optics which are known to pass it without much obstruction, they're going to behave like opposites. Imagine if the glass/plastic they used in the car blocked the light. Wouldn't really work.
There is a flip side to this though. Quick searches show that the safety of being absorbed and then dissipated by the water in the eye also makes that wavelength perform worse in rain and fog. I think a scarier concept is a laser that can penetrate through water (remember humans are mostly bags of salt water) which could, maybe, potentially, cause bad effects.
Depends on the wavelength of lidar. Near IR lidars (850 nm to 940 nm, like Ouster, Waymo, Hesai) will be focused to your retina whereas 1550 nm lidars (like Luminar, Seyond) will not be focused and have trouble penetrating water, but they are a lot more powerful so they instead heat up your cornea. To quote my other comment [1]:
> If you have many lidars around, the beams from each 905 nm lidar will be focused to a different spot on your retina, and you are no worse off than if there was a single lidar. But if there are many 1550 nm lidars around, their beams will have a cumulative effect at heating up your cornea, potentially exceeding the safety threshold.
Follow up question that you might know: would multiple LIDAR sensor actually be additive like that? If you can stand a foot away from a car's LIDAR sensor and be unharmed, then can't you have:
x^2 sensors at x feet from you and have the same total energy delivered? If sensors are actually safe to look at from 6in or 3in, then multiple the above table by 4 or 16.
It seems like, due to the inverse square law, the main issue is how close you can get your eye to a LIDAR sensor under normal operation, not how many sensors are scattered across the environment. The one exception I can think of is a car that puts multiple LIDAR arrays next to each other (within a foot or two). But maybe I'm misunderstanding something!
Do you if there has been any work how lasers affect other animals and insects?
Am I being catastrophically pessimistic to think that in addition to swatting insects as it moves forward, the cars lidar is blinding insects in a several hundred meter path ?
I’m very optimistic about automated cars being better than most humans but wonder about side effects.
If we have automated anti-mosquito vehicles just roaming around, the world would be a better place. There might be some second order effects from removing mosquitoes that we haven't predicted, but fuck mosquitoes.
Unfortunately not all insects are mosquitoes, and one reason we have many fewer birds in (e.g.) the UK than when I was young, is the decline of insect life.
GP is slightly wrong. IIRC those problematic LIDARs are operating at higher power than traditionally allowed, with the justification that the wavelength being used is significantly less efficient at damaging human eyes, therefore it's safe enough at those powers, which is likely true enough. But it turned out that camera lenses are generally more transparent than our eyes and therefore the justification don't apply to them.
Amusingly the lenses are worse than silicon at transmitting that wavelength.
1550nm might be worse for sensors because a good portion of the light is only being dumped into the metal layers - pure silicon is mostly transparent to 1550nm. Not sure how doped silicon would work. I can tell you that 1070nm barely works on an IQ3 Achromatic back…
A point source in the visual field will create a point image on the retina. The "sensor area" you're referring is what's necessary to capture the entire visual field simultaneously.
I looked this up for a laser-based projector, Class 2 is "blink reflex should protect you" and "don't be a doofus and stare into it for a long time". Look up the classifications on the google and you'll see other things like "don't look into the rays with a set of binoculars" and stuff.
Class 1 is pretty darned safe, but if you're continually bathed by 50 passing cars an hour while walking on a sidewalk... pitch it to a PhD student you know as something they should find or run a study on.
I have a Class 1 Makita (green) laser level, wide strong beam, excellent tool for landscaping. I accidentally looked into it from a 10 cm distance. It did not leave permanent damage, but for a few days I had the dot burned on my retina. And yes, I almost immediately closed my eye - within a few hundred ms's.
What about if you're walking or biking next to congested motorway and most of the vehicles have LiDAR running at the same time? That's a lot of photons.
This is inconsistent with the basic concept. It’s projecting and reading lasers . By default some emissions will hit people in the eye. Even invisible light can damage tissue , especially in the eye
Depends on the type of LIDAR. LIDAR rated for vehicle use is at a wavelength opaque to the eyes so it hits the surface and fluid of your eye and reflects back rather than going through to your cones and rods.
It isn't however opaque for optical glass (since the LIDAR has to shine through optical glass in the first place) so it hits your camera lens, goes straight through, and slams the sensor.
You seem to be implying that all automotive lidar are 1550 nm but that's not true. While there are lots of 1550 nm automotive lidars (Luminar on Volvo, Seyond on NIO) there are also plenty of 850 nm to 940 nm lidars are used in cars (Hesai, Robosense, etc). Those can pass through water and get focused to your retina, but they are also a lot lower power so they do not damage cameras.
Also although that energy longer than 1400nm is generally absorbed by the cornea and lens, it is still energy, and it is not a hard bandpass filter per se. Safety is relative at higher wattages.
NGL I thought sub 1550nm LIDAR had been banned for use in new automotive applications already? I clearly must be mistaken but I had thought that was the case.
Not banned. In addition to the Chinese lidars I mentioned, the Valeo Scala on Audi cars is 905 nm, and then there are also Ouster (865 nm), Innoviz (905 nm), Livox (905 nm) etc. The large spinning lidar on top of the Waymo Jaguar I-Pace is also purportedly 905 nm, although in the past they also had a swivelling 1550 nm lidar in the dome of the Chrysler Pacifica cars (situated just underneath a smaller spinning 905 nm one).
The eye safety threshold for 850/905 nm is a lot lower than 1550 nm, so they output way less power, but the much better sensitivity of silicon sensors makes up for it partially. You can also squeeze out more range using clever signal processing and a large optical aperture (which allows you to output more light, but since the light is spread out across the aperture, the intensity doesn't exceed the threshold). Typically, the range of 850/905 nm lidars is less than that of 1550 nm lidars though.
On the bright side, due to lower power, there hasn't been any instances (to my knowledge) of 850 nm and 905 nm lidars damaging cameras, whereas at least two different 1550 nm lidars have been known to destroy cameras (Luminar and AEye).
On the Luminar lidar website [1] they proudly advertise "1,000,000x pulse energy of 905nm".
I watched the livestream and they said their hardware is "Camera Safe". I am not sure if camera safe and eye safe are correlated, but I would hope/expect that they would not release something that isn't known to be eye safe. I guess it's possible that the long term effects could prove bad, and we will all end up getting "Lidar Eye" dead spots in our vision.
Digital camera sensors are much more sensitive than eyeballs, so it's not out of the realm of possibility that it won't leave a permanent line across your eyeball like it can to a camera sensor
When it becomes a widespread issue they'll just release Meta Glasses 5/Apple Vision 3 with the appropriate eye protection, and vision will be very affordable.
To date I can't think of any existing lasers which you are intended to look at during daily use. Most consumer facing lasers are either class 1 but hidden (CD-ROM), or class 2 but basically not shined into your eye (barcode reader).
The lack of accessible certification/testing docs for the lidars is also worrying. Where is the proof that it was even tested? Was it tested just via simulation, via a dummy eye stand-in, or with a real biological substitute?
What if there are biological concerns other than simply peak power involved with shining NIR into the eye? (For instance, it seems deep red light has some (beneficial) biological effects on mitochondria. How do we know that a pulsed NIR laser won't have similar but negative effects, even if it doesn't burn a hole in your retina.)
> class 2 but basically not shined into your eye (barcode reader)
TIL barcode readers can cause serious eye damage when misused. You'd think they'd have warning labels on these things given how common self-checkout is nowadays and how curious/stupid children can be
They're not always class 2, some are class 1. I'd maybe suspect the self-checkout ones are class 1 since the object is pressed right against it, while handheld barcode scanners are class 2.
If it's like everything else, it's "move fast, break things". I'm sure if it turns out to be harmful, we'll find out decades before regulation catches up
In terms of plain wattage, it cannot be dangerous. Unless, of course, you were to stand with your eye up against the sensor and maybe stare at it for a few minutes.
Apple doesn't own any foundries, so no. It's not trivial to spin up a DRAM foundry either. I do wonder if we'll see TSMC enter the market though. Maybe under pressure from Apple or nvidia...
I’ve been building this solo for about a year, so it’s still early but we’re starting to see real traction from creators testing it. More examples coming soon.
