Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

How do Tesla cameras self clean?

This site may earn commission on affiliate links.
My b-pillars are also fine. Trust me when I say its a non-issue other than the BUC (or radar). Only a few people have any issues and nothing is perfect. But its a non-issue (and I'm quite clear when something is an issue).

I've driven almost 10k winter miles through many (maybe 10 now) big snowstorms and I've had 2 radar failures and both were for 10-15 miles (10-15 minutes). Was it inconvenient? Yes. Was I prepared to take over -- yes because the conditions weren't great I was already giving the system a short leash so it was fairly seamless once the emergency klaxons started firing.

It really isn't an issue.

@mknox -- It won't be an issue. Even if the BUC is splashed the water clears and it can see (not that BUC is particularly useful (and its currently unused for AP)).

The repeaters are protected and heated and B-pillars mostly stay clean (they are enclosed in glass and apparently heated though very weakly).

I haven't had a chance to wash my car in 3 weeks (Its been scary cold). Even my BUC is harder to see out of but not completely blocked or useless. Just maintain and it'll be fine. Even in this extreme situation its fine. I don't doubt some catastrophic incident might occur, but nothing is infallible, even Elon.
 
The side cameras seem to be positioned in places that naturally stays clean (there will be dust, but not buildup that interferes with operation).

The rear camera seems to be the only one that consistently has problems. I wonder how the Model 3 one compares.
 
Well at least the BUC does get very dirty even after one drive in the rain, snow, slush, and worst in the salt. The worst part about salt is that it dries white onto the camera. If it is water it just leaves behind a ring you can wipe off later.
 
I agree the repeater and B pillar cameras are less likely to have problems

I don't agree the situation is as clear with the radar or the BUC to universally say "It really isn't an issue", though. First of all, the BUC requires near-daily cleaning for me in winter time. Not weekly. Already did in Model S days. And in bad weather even outside of winter. It is just a get-dirty design... This is not abnormal, mind you, rear cameras get dirty in general.

Second of all, the radar design is obviously problematic - the plastic in front of it is not heated (and too distant from the radar to benefit from its own heat), but does gather snow while driving. That's a design flaw IMO, pure and simple, I've never had such a easily blocked radar in any of my cars - and this I believe is my 15th automotive radar. Clearly most problematic.

It may well be that these issues manifest themselves differently regionally (different climates, different road clearing methods, different garaging cultures etc.). I wouldn't presume to tell anyone what their conditions are, just reporting my experiences.

The rest of the cameras have not been used for anything yet, so hard to say, beyond watching online snapshots and looking at them externally. But speculatedly they may very well fare a lot better.
 
I'm not quite sure why people believe that Tesla doesn't consider these issues during the design phase...

I'm sure they do. What we are pondering is how far will their prioritizations and choices succeed in real-life.

As said, Elon himself raised the prospect of a coast-to-coast driverless Summon.

This stuff matters in that scenario. It is a legitimate topic for pondering.

As for present-day issues, it seems they didn't ponder quite enough at least in the case of the backup cam and the front radar heating?
 
  • Like
Reactions: Demonofelru
It sounds like once it becomes a bigger problem (especially backup camera) someone can retrofit something to clean?
Hardware update to 2.5 or 3....unfortunately. I'm not expecting full FSD on the current hardware version now anyway so I will be expecting updates and I'm sure they will continue to improve the hardware in multiple ways. Addressing dirt, salt and mud is just one area.
 
  • Like
Reactions: 1 person
Maybe it's just me, but there seems to be this false premise being advanced that autonomous vehicles must be able to drive during extreme weather conditions.

I say false premise because even when a snow storm is predicted, many state governments tell motorists to stay off the roads so that road crews can do the job of clearing the snow from the roadways for good reason. It makes no sense to hop into an autonomous vehicle and expect it to drive through a foot of snow during blizzard conditions just as it makes no sense to drive your own car in this kind of weather.

Now to the question by the OPS... Others have done a good job of explaining how the cameras are protected, heated(in some cases) and cleaned (via wipers for front cams) .

My take, if conditions as so bad as to blind the cameras, freeze the sonic sensors, distort the radar then one should think twice about driving whether by self or autonomy. (and, like all machines, humans must still take responsibility for their maintenance and safe operation.)
 
  • Like
Reactions: robby and croman
It makes no sense to hop into an autonomous vehicle and expect it to drive through a foot of snow during blizzard conditions just as it makes no sense to drive your own car in this kind of weather.
I get your point, but I don't think it's a score.

The problem with the BUC (yes, in pre-AP, AP1, AP2, AP2.0, MS, MX) is that it gets clogged up even if there is no blizzard. Even if there is no rain storm. Look at this picture, for example:

BerylSpring_09-2-22.jpg


No blizzard, just plain, common, 'powdery' snow on the road. This stuff can blast the BUC like crazy. And it's a road condition many people drive on for months. Of course there's a billion variations too. Plus - it could be snowing, lightly or heavily, without it being 'blizzard' conditions. A super clean BUC is usually useless after just minutes of driving.

A wet, dirty tarmac can be just as bad, esp. if your driving above 40 or 60. The dirty water renders BUC useless.

Don't get me wrong, I love my BUC. When the conditions are right, I have it as a standard window. And as @AnxietyRanger pointed out, the clogging up is probably not a unique problem for teslas.

But you can't deny that it has the potential to be a significant problem for AV's if they're supposed to make use of the BUC in the sensing. It's not necessarily a safety problem (it could be, though), but it's certainly a very limiting factor if one has to get out of the car and clean it all the time to be able to drive autonomously.

That said, I don't think we'll ever see any new HW for automatic cleaning. Tesla will work around this somehow
 
It makes no sense to hop into an autonomous vehicle and expect it to drive through a foot of snow during blizzard conditions just as it makes no sense to drive your own car in this kind of weather.

My take, if conditions as so bad as to blind the cameras, freeze the sonic sensors, distort the radar then one should think twice about driving whether by self or autonomy. (and, like all machines, humans must still take responsibility for their maintenance and safe operation.)

I fear your's is the false premise. Driverless autonomous cars will not be able to choose what kind of weather conditions will meet them while on the road. Again, Elon's example was Summon your car from the other side of the country. At the very least a driverless car doing this will have to be able to handle leaving the road safely in most conditions.

Also, I fear your are probably wrong that it would take "a foot of snow during blizzard" to disable a suite like AP2/2.5. As things stand, the backup cam can be disabled by a splash of dirty water, the front radar by minimal snow if it has the right composition to get stuck on the front of the car.

These are very normal conditions in much of the world.

Now, nobody is claiming driverless cars don't need human maintenance. The question is, how will they cope with the basic driving task in varying conditions in real-life, between reasonable human attention. That is an interesting question.
 
  • Like
Reactions: Demonofelru
I think the problem with this thread is that some are thinking small or defensively about the current hardware suite, while others are pondering autonomous sensors and their autonomous maintenance on a larger or higher level. There is a discrepancy between many of us on what we're even talking about...

Again, let's return to Elon saying we can Summon our Tesla fron the other side of the country. Summon means driverless. When and how could this happen?

Retro-fitting cleaning to existing cars is going to be difficult... But they already took your money, so I guess they will have to.

First of all, Tesla obviously will not change the cleaning hardware on AP2 or AP2.5 cars after the fact. They are what they are and we shall see what they can be made to do. (I guess in absolute theory I might accept Tesla could, say, address the front radar heating or some such for FSD buyers retroactively were it absolutely mandatory and a plug-and-play swap, or other such simple sensor assembly swap, similar to the upgrading the AP computer, but I don't think even that is likely.)

Second, I don't think most of us think Tesla's full self-driving hardware, as it currently stands, will be able to handle the tougher weather environments - or perhaps be able to operate fully driverless at all beyond some geo-fenced California sun lab. That said, Tesla did say Level 5 capable hardware. That would mean being able to operate driverless in basically all conditions (Level 5 means you can remove the steering wheel), unlike Level 4 which would be able to do it in limited conditions, but not all. I think most of us just don't believe the Level 5 claim was true. It was usual Tesla hype and in the end we're likeliest to get less.

Third, the suite simply lacks the hardware redundancy in basically all areas but the front, which pretty much guarantees it will be limited. Lacking redundancy, it will be very easily disabled by one unfortunate spray of mud, it may even prove the camera placements are inadequate for too many scenarios. Compared to competing future suites that we know of publicly being worked on with 360 radar plus additional lidar coverage as well as the cameras, AP2/AP2.5 cars are equipped thinly. (Of course a reasonable expectation is that Tesla too has private development of a more robust suite.)

All this comes down to this: IMO what we have in AP2/2.5 is Tesla's bare minimum what they feel they need to progress their self-driving efforts on the road. It is not an optimal suite by any means, it is the absolute minimum suite Tesla needed to start feeding their monster. It probably is capable of fair-weather urban California cruising in big sun, on big roads. So no, we won't be Summoning AP2/2.5 cars across countries. Perhaps in theory Tesla might geo-fence something in the U.S., but RoW, we're better off forgetting about that.

BUT, BUT, BUT. Many manufacturers don't have cleaning for their cameras, because most cameras are not yet used for autonomous functions yet and can be expected to be cleaned by humans. Not even Tesla's suite is really used yet, so not even Tesla's interest in making it more robust is high yet. As said, Tesla just needs some hardware on the road to feed their monster. However, it remains an interesting question where do things go from here. The Germans already use well positioned water sprays to clean some cameras and Waymo has used innovative camera/lidar wipers. Down the road this will become more and more important.

I'm pretty confident that especially once we're at fair-weather, warm-weather autonomy, car manufacturers will be very interested in solving the fair-weather, cold-weather autonomy (indeed Waymo is already testing that) and then the bad-weather autonomy and finally the terrible weather autonomy, and the latter three will necessitate great autonomous cleaning solutions and redundancy for the sensors...

I also expect big things from Volvo. Their development lab is very different from California or even the Germans. Let's not forget Volvo's innovative headlamp wipers back in the day... our California cars don't even have headlamp washers. :)

It is absolutely certain autonomous car progress will not stop at fair weather. It may start there, but it will not end there.

And IMO that's a very legitimate topic of conversation, what will those autonomous sensor cleaning solutions be?
 
Last edited:
For FSD to work the cameras must have some way to clean themselves. It's clearly no good if your car gets splashed with mud or simple build up of dust/soot and can't self drive any more, stranding it far from where you are and without any means to recover.

Obviously the front camera has the windscreen wipers. What about the others, how do they recover from dirt?
autonomous car drives into autonomous car wash, done.
 
  • Funny
Reactions: daktari
Doesn't really help if it gets splashed or covered with snow/salt while driving, then has to stop in the middle of the road because it can't cope with multiple camera failures.

Your assumption is that all the cameras must be functional for the vehicle to potentially find a solution, (e.x. drive to a car wash). It is more than likely that the front cameras and radar are capable of handling the majority of driving tasks, especially if the vehicle need only "limp" to a solution.
 
Your assumption is that all the cameras must be functional for the vehicle to potentially find a solution, (e.x. drive to a car wash). It is more than likely that the front cameras and radar are capable of handling the majority of driving tasks, especially if the vehicle need only "limp" to a solution.

You are assuming it will be capable of operating with one or more cameras not working.

Also, they need to build a vast network of automatic car washes, because the ones designed for humans won't work. The car can't hand over cash or enter an activation code, for example.
 
You are assuming it will be capable of operating with one or more cameras not working.

Also, they need to build a vast network of automatic car washes, because the ones designed for humans won't work. The car can't hand over cash or enter an activation code, for example.

It's not really an assumption. Current AP 2.0/2.5 doesn't not use all the cameras.

Right, but we're discussing hypotheticals, are we not?