Digital Vision Safety

[Captain Ahab]

Folks, please chime in on this idea:

Some digital cameras are able to image light in part of the infrared specturm (try shooting a beam from your remote control at your cell phone's camera).

Because cameras are able to dectect infrared light, why not make infrared as a part of the solutions to driver assistance and safety features?

The prosals are nothing new, but I believe these applications may be not utilized at this point.

My specific proposals are:
A. Passive infrared emission by way of chemical doping of automotive paint; OR
B. The use of active infrared emitters, like tail lights; OR
C. The application of infrared reflective tape on all sides of vehicles (such that turned over vehicles are visible); OR
D. Applying infrared imaging as part of the computer imaging based driving algorithim (no hardware changes).

Do any of these ideas have merit?

I do not use social media, maybe this merits exposure to a larger community to chime in on the concept?

 

[DanCar]

What is the advantage of emitting infrared, when you can emit visible light via headlights?

 

[Captain Ahab]

My thoughts are that infrared reflection or emission can provide additional digital imaging information, that will have minimal to zero impact on the appearance of the vehicle in the visible spectrum.

Furthermore, I believe at infrared wavelengths, the radiation of these energies penetrate atmospheric and weather phenemenon better than visible light can (just as the sky transitions from blue to orange as the sun goes down).

 

[Captain Ahab]

My thoughts are that because infrared emissions can be:
Cheap
Effective
Invisible to our eyes

My theory is that because of the above, the consumer acceptance of the technology might be higher.

I am thinking that trailer roofs and sides could have the IR emissions, thus preventing collision, when a sky or road colored, overturned trailer or cargo occupies the intended path of the computer vehicle functions.

Safety is the ultimate goal here.

 

[Quamera]

IR sensitive cameras would also make it easier to detect people and animals near the road because of their heat signature even if they were not moving. I assume FSD can detect and avoid animals and people crossing a road. Does anyone know how effective it is and at what size does it ignore a live animal, a mouse, rat, rabbit.

 

[nvx1977]

My thoughts are that infrared reflection or emission can provide additional digital imaging information, that will have minimal to zero impact on the appearance of the vehicle in the visible spectrum.

Furthermore, I believe at infrared wavelengths, the radiation of these energies penetrate atmospheric and weather phenemenon better than visible light can (just as the sky transitions from blue to orange as the sun goes down).

There's a difference between the IR from a remote and the IR for thermal imaging (FLIR). Most digital camera sensors can pick up a bit of IR and a bit of UV (basically a bit outside of the human-visible light spectrum).

But to get the advantages you talk of where an IR image is sufficiently different from visible light imagery and is useful for self-driving, you're talking about much longer wavelength IR, and that requires much more expensive sensors.

From Tesla's POV, it has to be cost- and energy-efficient. The component cost of a cell phone camera is around $25. An equivalent FLIR camera, like those you can attach to your phone is about 10x the cost. Something that is automotive grade will be much more expensive.

I'm no expert on FLIR imagery, but I would imagine one of the perceived benefits would be detecting heat signatures for vulnerable road users (VRUs). I would be concerned that during summer, there would be lots of heat noise from pretty much everything in the environment.

So maybe it's viable as a layer of redundancy, but only if it's cheaper. My guess is that if it were viable, more entities in the industry would be experimenting with it already. Sandy Munro is a fan of using FLIR for self-driving applications.

 

[Daniel in SD]

What about putting an infrared laser on the car that scans across the field of view? Then you can measure how long it takes for the reflected light to come back to determine where things are.
EDIT: This would have the additional advantage of being able to detect pedestrians whose infrared emitters have run out of batteries.

 

[JHCCAZ]

What about putting an infrared laser on the car that scans across the field of view? Then you can measure how long it takes for the reflected light to come back to determine where things are...

That's called Lidar.

 

[JHCCAZ]

What is the advantage of emitting infrared, when you can emit visible light via headlights?

One advantage is that the beam pattern can be very broad, like High beams but without blinding other road users. It's not a replacement for visible spectrum lighting but a reasonable enhancement.

 

[Daniel in SD]

One advantage is that the beam pattern can be very broad, like High beams but without blinding other road users. It's not a replacement for visible spectrum lighting but a reasonable enhancement.

You know what also has a very wide beam pattern?

 

[Dewg]

Two thoughts: If all the cars start using IR blasters, lasers, radar, etc., will the roads just get flooded with these emissions and how will your car know it's picking up a return from its beam emitter and not a reflection from another vehicle? I know energy can be modulated, so perhaps your car sends out a specific frequency and it's looking for a return at that specific frequency. But it's going to be hard if 20-30 cars at a large intersection are all sending out lasers/radar, bouncing off every object and building around.

Can't someone build a jamming device that floods an area with broad-spectrum energy, causing all these L3-5 cars to fail? Like radar jammers many years ago could stop the police from reading your speed - only this time your car fails to drive. Since LIDAR, RADAR, and IR are all invisible to human eyes, we wouldn't know it was happening.

Visual scanners, on the other hand, can't be as easily jammed without the driver seeing where the jamming was taking place - such as someone high-beaming to temporarily blind you. It's a little sci-fi right now, but I can totally see a 12-year old with a jammer out there trying to cause driverless cars to crash.

 

[Daniel in SD]

Two thoughts: If all the cars start using IR blasters, lasers, radar, etc., will the roads just get flooded with these emissions and how will your car know it's picking up a return from its beam emitter and not a reflection from another vehicle? I know energy can be modulated, so perhaps your car sends out a specific frequency and it's looking for a return at that specific frequency. But it's going to be hard if 20-30 cars at a large intersection are all sending out lasers/radar, bouncing off every object and building around.

Can't someone build a jamming device that floods an area with broad-spectrum energy, causing all these L3-5 cars to fail? Like radar jammers many years ago could stop the police from reading your speed - only this time your car fails to drive. Since LIDAR, RADAR, and IR are all invisible to human eyes, we wouldn't know it was happening.

Visual scanners, on the other hand, can't be as easily jammed without the driver seeing where the jamming was taking place - such as someone high-beaming to temporarily blind you. It's a little sci-fi right now, but I can totally see a 12-year old with a jammer out there trying to cause driverless cars to crash.

These work well against all pneumatic tire vehicles:

Assault rifles with large capacity magazines and auto sear devices are widely available in the US and work well against non armored vehicles.

Seriously though, IR flashes can be made short enough that they don't interfere with each other. Most cars have RADAR now and it works fine because the pulse widths are very short.

 

[Daniel in SD]

Maybe LIDAR should be rebranded as a "time domain IR imager"

 

[Dewg]

Seriously though, IR flashes can be made short enough that they don't interfere with each other. Most cars have RADAR now and it works fine because the pulse widths are very short.

Ah, but those systems are for emergency and narrow use parameters with a human driver. Take away the human, and purely rely on them at high speed, and their focal length and power will need to be greatly increased, no? Instead of a few hundred feet, they need a few thousand feet.

 

[Quamera]

Has anyone had any experience with FSD dodging or avoiding wildlife that may wander or dart onto the road? Is a Tesla any good at it and does it use the full dynamic abilities of the vehicle without risking a loss of control.

 

[Daniel in SD]

Ah, but those systems are for emergency and narrow use parameters with a human driver. Take away the human, and purely rely on them at high speed, and their focal length and power will need to be greatly increased, no? Instead of a few hundred feet, they need a few thousand feet.

Here's an EETimes article about it. It seems to be a concern but there's not a lot of hard data which is odd because ~100% of new cars have RADAR.

“Radar-to-radar interference is a still unknown and as applied researchers who work with radar almost every day VSI cannot say we have ever experienced radar-to-radar interference from another vehicle while testing on public roads,” said VSI Labs’ Magney. “We can assume that we are exposed because so many vehicles on the road today have a mix of radars for short ranging to long ranging,” he added.

https://www.eetimes.com/radar-safety-warning-dont-cross-the-streams/

 

[JHCCAZ]

...there's not a lot of hard data which is odd because ~100%* of new cars have RADAR.

* ((ALL_UNITS - TESLA_UNITS) / ALL_UNITS)×100%