Dashcams at risk from autonomous vehicles!

[Nigel]

Are our dashcams safe?

A man attending this week's CES show in Las Vegas says that a lidar sensor from startup AEye has permanently damaged the sensor on his $1,998 Sony camera. Earlier this week, Jit Ray Chowdhury, an autonomous vehicle engineer at the startup Ridecell, snapped photos of a car at CES with AEye's lidar units on top. He discovered that every subsequent picture he took was marred by two bright purple spots, with horizontal and vertical lines emanating from them.

AEye CEO Luis Dussan stressed that AEye lidars pose no danger to human eyes in an email to Arstechnica. But he didn't deny that AEye's lidars can cause damage to camera sensors. AEye has offered to buy a new camera instead of the damaged one.

https://arstechnica.com/cars/2019/0...r-was-so-powerful-it-wrecked-his-1998-camera/

 

[tempviewer12]

I doubt it. Remember the regulations for car makers in this area are not yet well defined. Lidar systems are currently cost prohibitive and not used for any production vehicles. If Lidar was to become more popular, it wouldn't just be our dashcams in trouble, it would be the police's cameras, ANPR cameras, speed cameras, CCTV... you name it. Without a doubt there will be some regulation limiting the output spectrum or power or whatever to stop this from happening.

 

[Dashmellow]

I suppose it "may" be possible to fit camera sensors with 1550nm filters that could protect them from AEye type lidars without affecting the color balance of normal photography. On the other hand, even if effective, they could end up affecting a vast number of existing cameras in today's society as @tempviewer12 has already mentioned. Ironically, 1550nm lasers are touted as the "ideal wavelength of choice" as a selling point due to its retina-safe nature at high power.

 

[Nigel]

...not used for any production vehicles.

Over Christmas my brother demonstrated his Toyota's system by driving into me - totally ineffective!

Toyota Safety Sense's Pre-Collision System uses a camera and laser to detect other vehicles in the road ahead and will detect a potential rear-end collision. If you don't react in time, the Pre-Collision System kicks in to stop or limit the severity of an impact.

https://www.toyota.co.uk/world-of-toyota/safety/toyota-safety-sense

I suppose it "may" be possible to fit camera sensors with 1550nm filters that could protect them from AEye type lidars without affecting the color balance of normal photography. On the other hand, even if effective, they could end up affecting a vast number of existing cameras in today's society as @tempviewer12 has already mentioned. Ironically, 1550nm lasers are touted as the "ideal wavelength of choice" as a selling point due to its retina-safe nature at high power.

It may be retina-safe, but what about my skin, cloths, car paintwork etc, if it can burn a silicon sensor sufficiently that it no longer works correctly then it can damage other things as well!

 

[Dashmellow]

It may be retina-safe, but what about my skin, cloths, car paintwork etc, if it can burn a silicon sensor sufficiently that it no longer works correctly then it can damage other things as well!

1550nm laser diodes can indeed burn your skin as they are also used in cosmetic procedures but this requires close range, collimated focus and relatively prolonged periods of exposure. The amount of effect on your skin is also related to how much water exists within each cell hit with a collimated laser. Focus of a laser through lens optics is also at play when it comes to the damage to camera sensors. Lidar on autonomous cars is broadcast at greater distances and for minute fractions of a second exposure time during scanning and thus is far less likely to burn your skin or damage car paintwork, etc., certainly no more than the spectrum of light and radio frequencies we are all bathed in on any given day. I'm no expert in this but I believe that 1550nm is a highly specific choice for lidar and telecommunications purposes as greater wavelengths infrared absorption starts to increase dramatically to the point were it does require more attention to attenuation.

Still, the article you posted raises many interesting questions about this technology becoming ubiquitous in our society.

 

[Nigel]

but this requires close range

Given the statement in that article: "AEye says that its lidar has a range of 1,000 meters.", if the lasers can put a small dot of light on a vehicle 1Km away in order to measure its distance, then your "close range" is maybe a few meters?

I do wonder how they have proved that such a laser is "safe", given that if these vehicles become common place then if walking or driving along a busy road there may be several hundred vehicles within 1Km radius and many of us spend a significant proportion of our lives on the road, some of us will spend most of our working lives exposed to these things, as will our dashcams... I also wonder how they will cope with seeing each others laser beams when there are hundreds scanning each other within the 1Km radius, could be massive autonomous confusion!

 

[Dashmellow]

Given the statement in that article: "AEye says that its lidar has a range of 1,000 meters.", if the lasers can put a small dot of light on a vehicle 1Km away in order to measure its distance, then your "close range" is maybe a few meters?

I do wonder how they have proved that such a laser is "safe", given that if these vehicles become common place then if walking or driving along a busy road there may be several hundred vehicles within 1Km radius and many of us spend a significant proportion of our lives on the road, some of us will spend most of our working lives exposed to these things, as will our dashcams... I also wonder how they will cope with seeing each others laser beams when there are hundreds scanning each other within the 1Km radius, could be massive autonomous confusion!

The article also explains that the AEye lidar is a time-of-flight system using a short pulse of light and measuring the time it takes for it to bounce back. These scans occur in mere nano-seconds at very low power levels compared with medical lasers that are used "close range" (inches) or communications lasers at 1550nm or greater which are deployed in closed systems.

I agree though, this technology requires much more research and greater scrutiny about the effects of widespread constant exposure. Apparently, there is also concern about their accuracy as these systems have shown a tendency to become confused by various objects they encounter.

 

[Nigel]

If the power levels really are so "very low", how come it damaged a $2000 camera sensor?

 

[jokiin]

if there's any chance of these damaging cameras they'll get outlawed in no time, the government won't want anyone damaging all their $afety camera$ (revenue collecting) gear, there'$ too much at $take here

 

[Dashmellow]

If the power levels really are so "very low", how come it damaged a $2000 camera sensor?

Well, sensors are quite sensitive and prone to damage if the input exceeds their specs. For example, it is fairly common for bright direct sunlight to damage CCTV sensors. This happens when newbie installers don't know to avoid that.

Lidar systems don't particularly need to blast everything in sight with super-powerful beams, they just need a ping back to their sensors from whatever they happen to hit. As for the sensor in the camera, the laser damage to individual pixels was probably amplified by the fact that the photographer was focusing his camera directly at the beam. Also, perhaps the 1550 nm frequency is beyond the spec the sensor can handle when a collimated, lens focused laser hits it. This would be different than the nano-second of exposure your eyes or car paint would be subjected to.

The beam may have indeed damaged the camera sensor but I believe your fears about damaged skin, clothes and car paint are wildly exaggerated.

Like I've already said though, this will all require a lot more research to understand the possible pitfalls and dangers of the widespread introduction of this technology to society.

 

[Dashmellow]

if there's any chance of these damaging cameras they'll get outlawed in no time, the government won't want anyone damaging all their $afety camera$ (revenue collecting) gear, there'$ too much at $take here

As mentioned, a simple filter install "may" eliminate the possibility of damaging camera sensors.

 

[Nigel]

1550 nm is well into the infrared, I wouldn't expect much to make it through the standard IR cut filter, I would also expect it to be well out of focus if the lens was correctly focused for visible wavelengths.

 

[jokiin]

As mentioned, a simple filter install "may" eliminate the possibility of damaging camera sensors.

governments have to add filters to all their existing cameras, or they just legislate the offending products out of existence, tough choice

 

[Dashmellow]

governments have to add filters to all their existing cameras, or they just legislate the offending products out of existence, tough choice

I think it will depend upon how fast autonomous vehicle technology gets adopted. Perhaps some old municipal cameras and CCTV will get phased out as some others get modified with filters as the newer tech comes online.

Also, we still don't know the extent of the camera sensor problem. All we know is this one example with the Sony. It could be a matter of certain sensors being vulnerable to this issue more than others. The one in the camera that got damaged is a 42.4 megapixel full 35mm frame Exmor R.

 

[Dashmellow]

1550 nm is well into the infrared, I wouldn't expect much to make it through the standard IR cut filter, I would also expect it to be well out of focus if the lens was correctly focused for visible wavelengths.

Hard to say. The dichroic IR cut-off filters in our dash and action cams are confined to fairly narrow segments of the IR band. For example, the standard ones sold by Treeye on AliExpress are 650nm.

 

[Dashmellow]

If the power levels really are so "very low", how come it damaged a $2000 camera sensor?

I've been thinking further about your question and I don't really have a definitive answer but as far as power levels on the lidar device are concerned my hunch is that it wouldn't be too much different than your average hand held presentation laser pointer.

In other words, it could indeed project out for 1000 meters but it wouldn't burn the paint off a car or damage skin and unlike many of these types of consumer level lasers the 1550nm lidar won't harm the retina of an eyeball.

 

[Nigel]

I've been thinking further about your question and I don't really have a definitive answer but as far as power levels on the lidar device are concerned my hunch is that it wouldn't be too much different than your average hand held presentation laser pointer.

In other words, it could indeed project out for 1000 meters but it wouldn't burn the paint off a car or damage skin and unlike many of these types of consumer level lasers the 1550nm lidar won't harm the retina of an eyeball.

During the last solar eclipse a few cameras were damaged by the sun, but it was only the lenses that got damaged, the sensors where OK; if they can cope with the power of the sun then I suspect these lidars are rather more powerful than you are suggesting. Remember they have to detect the light returning off a car that is painted matt black, otherwise there will be a lot of accidents, and they claim to be able to do it at a return distance of 2Km!

 

[Nigel]

Ahh, the normal ones use 200 watt lasers, supposedly in very short pulses, wouldn't do for it to get stuck on, that should be enough power to cut through the steel used in car bodies!

The refined requirement then becomes 100 to 200 W transmitted as a pulse of 1 to 2 ns at a 1- to 2-MHz repetition rate.

https://www.electronicdesign.com/power/gan-devices-power-next-generation-lidar-systems

However these long range ones are 40 times more powerful:

The key to that success is looser safety rules that let Luminar’s lasers fire pulses 40 times more powerful at 1550 nm than 905-nm lasers are allowed to fire in a conventional car.

https://spectrum.ieee.org/cars-that...g/under-the-hood-of-luminars-long-reach-lidar

So that is pulses of 200 * 40 = 8KW
Now I understand why the sensors get damaged!

 

[Dashmellow]

During the last solar eclipse a few cameras were damaged by the sun, but it was only the lenses that got damaged, the sensors where OK; if they can cope with the power of the sun then I suspect these lidars are rather more powerful than you are suggesting. Remember they have to detect the light returning off a car that is painted matt black, otherwise there will be a lot of accidents, and they claim to be able to do it at a return distance of 2Km!

During the last solar eclipse a few cameras were damaged by the sun, but it was only the lenses that got damaged, the sensors where OK; if they can cope with the power of the sun then I suspect these lidars are rather more powerful than you are suggesting. Remember they have to detect the light returning off a car that is painted matt black, otherwise there will be a lot of accidents, and they claim to be able to do it at a return distance of 2Km!

Damaged sensors from direct sunlight exposure are a fairly common problem with some CCTV installations and manufacturers always offer cautionary installation warnings as a result. As for matte black cars there is a difference between the visible light spectrum and 1550nm infra-red, so I'm not sure power levels ultimately translate into how well lidar can see them. Obviously, you too don't have definitive answers.

The articles you reference are very interesting. Good find! No mention of burnt skin, clothing catching on fire or scorched car paint though.

 

[SawMaster]

After paying for hundreds or thousands of lawsuits, the technology will become self-regulating. If not, the governments will regulate it and as they usually do, each location will have different regulations which may restrict the car to being used in only certain Nations or States. As I see it, there is no valid reason to have resolution at 1000m when half that is sufficient at legal driving speeds. If they can't make the idea safe for everyone and their belongings, then the idea needs to be better developed before considering it's use, or another idea used in it's place.

Phil