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

Tesla Software updates - Australia

This site may earn commission on affiliate links.
Now that I have been on 2024.8.9 for a while, I can say the wipers on average are much better than earlier versions this year, however there are occasional random auto wiper decision issues. On some days and drives they are near perfect, on others they are way too slow during heavier rain or way too fast with barely any rain. Most of the last week they were fine with few exceptions where I needed to manually trigger them. Yesterday they were near perfect in the morning when it was raining, however in the evening when there was no rain and just a little mist pulled up from wet road surface they were going full tilt, so I had to manually change them to setting "I" for rather slow intermittent wipe. This morning I washed the car window before going out and it was lovely sunny morning, with no rain at all, First 5 minutes of drive all was well, then I stopped at red light and suddenly my wipers started and kept going for about a minute until I manually turned them off. There was not dirt on window and no visible reason for wipers to turn on. They still have long way to go to get that fully stable.
 
Now that I have been on 2024.8.9 for a while, I can say the wipers on average are much better than earlier versions this year, however there are occasional random auto wiper decision issues. On some days and drives they are near perfect, on others they are way too slow during heavier rain or way too fast with barely any rain. Most of the last week they were fine with few exceptions where I needed to manually trigger them. Yesterday they were near perfect in the morning when it was raining, however in the evening when there was no rain and just a little mist pulled up from wet road surface they were going full tilt, so I had to manually change them to setting "I" for rather slow intermittent wipe. This morning I washed the car window before going out and it was lovely sunny morning, with no rain at all, First 5 minutes of drive all was well, then I stopped at red light and suddenly my wipers started and kept going for about a minute until I manually turned them off. There was not dirt on window and no visible reason for wipers to turn on. They still have long way to go to get that fully stable.
Yeah, I still get the odd 'dry wipe' too.
 
I really don't want to be a party pooper. But...

It might be time to remind everyone here that unless you see physical evidence of changes on a screen, or the release notes specifically state that they've addressed an issue (such as poorly functioning auto wipers), it is highly unlikely that anything has changed under the hood, and your observations are simply repeat observations of the random functioning (or not) of the feature in question.

We get tons of updates in Australia that are not applicable to our specific region/model/year/hardware version. But we get them, because that is the strategy of Tesla's continuous software integration scheme, and gives them real-life testing that they didn't break anything in a release targeted at the US or European markets. Remember we are a vanishingly small and economically uninteresting market for Tesla.

A note on the autowiper or autopilot behavours specifically:
AI driven activation does not work in the same way as a detector that measures the state of a physical quantity. Traditional rain detectors work by measuring changes in the reflected IR beam sent out through your windscreen by the rain detector affixed on the inside. If the reflectivity of that beam changes, there's probably water (or other IR refractive or dispersive material) on the other side of the windscreen. Which is why everybody else's autowipers work better, because there's an incredibly high correlation between a change of the reflected IR when water is on the other side.

The cameras used for rain detection in our Teslas don't work like that. The neural network trained on what rain on the windscreen looks like might be catching a sunlight glint off a speck of dust, or a fly, or any other object in front of the camera, and the neural net produces an output that says "this might be rain" and therefore they turn on. But the *specificity* of this detection is much lower than for an IR beam. Hence the crappy performance of our auto wipers.
 
I really don't want to be a party pooper. But...

It might be time to remind everyone here that unless you see physical evidence of changes on a screen, or the release notes specifically state that they've addressed an issue (such as poorly functioning auto wipers), it is highly unlikely that anything has changed under the hood, and your observations are simply repeat observations of the random functioning (or not) of the feature in question.

We get tons of updates in Australia that are not applicable to our specific region/model/year/hardware version. But we get them, because that is the strategy of Tesla's continuous software integration scheme, and gives them real-life testing that they didn't break anything in a release targeted at the US or European markets. Remember we are a vanishingly small and economically uninteresting market for Tesla.

A note on the autowiper or autopilot behavours specifically:
AI driven activation does not work in the same way as a detector that measures the state of a physical quantity. Traditional rain detectors work by measuring changes in the reflected IR beam sent out through your windscreen by the rain detector affixed on the inside. If the reflectivity of that beam changes, there's probably water (or other IR refractive or dispersive material) on the other side of the windscreen. Which is why everybody else's autowipers work better, because there's an incredibly high correlation between a change of the reflected IR when water is on the other side.

The cameras used for rain detection in our Teslas don't work like that. The neural network trained on what rain on the windscreen looks like might be catching a sunlight glint off a speck of dust, or a fly, or any other object in front of the camera, and the neural net produces an output that says "this might be rain" and therefore they turn on. But the *specificity* of this detection is much lower than for an IR beam. Hence the crappy performance of our auto wipers.
That's it... I'm selling mine :p;)
 
I really don't want to be a party pooper. But...

It might be time to remind everyone here that unless you see physical evidence of changes on a screen, or the release notes specifically state that they've addressed an issue (such as poorly functioning auto wipers), it is highly unlikely that anything has changed under the hood, and your observations are simply repeat observations of the random functioning (or not) of the feature in question.

We get tons of updates in Australia that are not applicable to our specific region/model/year/hardware version. But we get them, because that is the strategy of Tesla's continuous software integration scheme, and gives them real-life testing that they didn't break anything in a release targeted at the US or European markets. Remember we are a vanishingly small and economically uninteresting market for Tesla.

A note on the autowiper or autopilot behavours specifically:
AI driven activation does not work in the same way as a detector that measures the state of a physical quantity. Traditional rain detectors work by measuring changes in the reflected IR beam sent out through your windscreen by the rain detector affixed on the inside. If the reflectivity of that beam changes, there's probably water (or other IR refractive or dispersive material) on the other side of the windscreen. Which is why everybody else's autowipers work better, because there's an incredibly high correlation between a change of the reflected IR when water is on the other side.

The cameras used for rain detection in our Teslas don't work like that. The neural network trained on what rain on the windscreen looks like might be catching a sunlight glint off a speck of dust, or a fly, or any other object in front of the camera, and the neural net produces an output that says "this might be rain" and therefore they turn on. But the *specificity* of this detection is much lower than for an IR beam. Hence the crappy performance of our auto wipers.
Agree. The wipers are still hit and miss. How does the front camera not recognise a sudden heavy downpour and need a manual nudge to get going
 
The cameras used for rain detection
I think I read somewhere a while ago that only 1 camera is currently used for rain detection, and that "actually auto-wipers (AAW?)" will be coming that will use more than 1 camera. I have seen no evidence that that has happened yet, so any observations of improvement are, as you say, likely just based on chaos theory.

Some people continue to get good results, others bad, others varied. It can vary geographically - for example, Sydney/Brisbane tends to either rain or not rain. Melbourne is just kind-of drizzling all the time (except when it's not) and we don't even notice until we can't see through the windscreen because the wipers aren't working.

There was a (noticed by many) change in behaviour a few months ago which seemed to reduce dry-wipes but also reduce wet-wipes. This may have been simply Tesla reacting to the dry-wipes complaints and winding down the sensitivity level without any actual improvement in the NN rain-detection intelligence. This lead some people to wonder that if they can do that, then why don't Tesla just (as an interim fix) allow a user-adjustable sensitivity level, which many cars with traditional rain-sensor auto-wipers have. However, it might not be that simple.

I'm hoping that now that Elon has tweeted that they are "no longer compute constrained" they can get a team working on the NN-rain-training and just fix the bloody problem, because it's one of those things that really detracts from the car, however "first-world-problem-ish" it may seem.

I don't know how many people have asked me how I like my Tesla, and my first response is always "the windscreen wipers suck". (I usually continue with "but otherwise I love it", but the fact that that is the first thing coming out of my mouth is saying something).
 
Try and look for the good things rather than focus on those that are not perfect. You won’t have to look hard as the Tesla is an incredible thing of beauty :)

TLDR:
"Yes. I can see those two bad bricks. But I can see the 998 good bricks as well."

 
When my TP user cohort forgets to replace an empty roll, a false wipe (using newspaper/magazine paper) can yield worse results than a no-wipe.

Same goes for the autowipers. Driving into low sun, a false wipe on an otherwise dry window has significant potential to cause streaks, making the windscreen into a bright and visually impenetrable screen of glare.
 
When my TP user cohort forgets to replace an empty roll, a false wipe (using newspaper/magazine paper) can yield worse results than a no-wipe.

Same goes for the autowipers. Driving into low sun, a false wipe on an otherwise dry window has significant potential to cause streaks, making the windscreen into a bright and visually impenetrable screen of glare.
And unlike the bathroom, the Tesla doesn’t have curtains nearby that can be used in a pinch :)
 
Tesla are light years ahead with these UIs.
Wouldn’t be hard… but Tesla are light years behind in true usable UI… they need a lesson or two from Apple!
Even with this improvements the UI is klunky, and items in sub menus often totally out of context. It’s always the same when the tech drives the UI instead of true user centred design.