There is still no penalty. Even If a company is found to have breached this law they cannot be penalised. The worst they get is forced to refund or replace. This means companies take the risk of seeing how far a consumer will go. If the law had a massive penalty for non-compliance I think we would see a lot more companies co-operating quickly
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Tesla Software updates - Australia
- Thread starter Grenadine
- Start date
The squeaky wheel gets the most grease. If no one is going to protect your interests, there are mechanisms to do it yourself.
I wish people would standup more and assert their consumer rights. Then less of this will occur
I once got Apple to replace a laptop logic board 2 years after end of warranty. They eventually replaced it under Consumer guarantee, not under warranty.
On version 23.26.7 the wipers are still painful. Now with this version, sometimes the wipers go flat out when simply activating the wiper wash function (clean windscreen BTW). It seems software alone simply won't fix the ongoing wiper issues that have plagued Tesla's for years now. Also, just returned from a road trip form Melbourne to Yamba and with regular light misty rain wipers still operated very chaotically. Wipers tend to be always over active compared to what is required. Constantly need to turn them off or settle for the slow intermittent option.
I think it will be the last EV road trip we will do. We have always chosen not to go during school holidays to avoid any possible charger mayhem, but even now (particularly in NSW) it's getting too busy with so many more Tesla's and other EV's on the road. As the weeks tick by and the EV count increases, the charging infrastructure can't keep up with the demand. School holiday charging wait times will increase more and it will be crazy for non Tesla's.
I'll go back to using the long range diesel SUV for road trips in future.
The Y can remain the run around vehicle for the 90% of driving within 200kms of home.
I think it will be the last EV road trip we will do. We have always chosen not to go during school holidays to avoid any possible charger mayhem, but even now (particularly in NSW) it's getting too busy with so many more Tesla's and other EV's on the road. As the weeks tick by and the EV count increases, the charging infrastructure can't keep up with the demand. School holiday charging wait times will increase more and it will be crazy for non Tesla's.
I'll go back to using the long range diesel SUV for road trips in future.
The Y can remain the run around vehicle for the 90% of driving within 200kms of home.
OzTzlaBloke
Member
there’s an upcoming update to revision 4 of the wiper systems. This is apparently the last neural net restricted to a single camera and the update will move the wipers onto the platform that uses them all for like the other neural nets
...and all we've got is Purple Rain instead, and only if Spotify/Apple Music doesn't lock up the MCU. My silly wipers are in the off position until I see rain.
On a different but related note, just drove Brisbane to Sydney yesterday on a roundabout detour route with 2023.26.8 and noticed a few interesting things:
Range prediction has been surprisingly spot on. That's definitely been getting better for the last several releases. Predicted arrival for a 350km away SC was at 1%. I wanted a little more buffer so I planted myself behind a B-double for about 100km and gained 5% for an arrival estimate of 6%. Arrived with 5% at the SC. This is all the more surprising as the last 250km of that drive were on the Grafton - Glen Innes - Tamworth route, a route I've never driven before and that has some massive ascents and descents (by Aussie standards). So Tesla would have no data of mine on that route. In fact on this entire ~2,700 km to Brisbane and surrounds the car overall has provided really accurate SOC predictions. As long as I let AP drive of course. This all goes (ludicrously) rapidly out the window as soon as I drive with play factor.
Speed limit sign recognition worked really well*, except for where it didn't: Had another two random instances of the car reading non-existent 50 signs in a 110. Even saved the video to go back and try and find which sign it misinterpreted. Turns out there weren't any signs visible at all. I can only assume that comes out of an overly imaginative database. Also encountered several really effing annoying repeat instances of the car imposing a speed limit of 80 when it was a sign "80 for trucks and buses". They still don't have compound signs sorted apparently.
Autosteer performed nearly flawlessly. Still swerves on country roads that widen for a turn-off, but really did well at all other times, even in situations where I was borderline confused myself on where to go it managed just fine. Brisbane has a few "interesting" locations with complex road layouts and turning options. AP and I saw those for the first time, and I can honestly say AP drove those better than me seeing them for the first time.
I don't often pay attention to the trip meter data, but on this trip I noticed a discrepancy between what Teslafi and the car report. According to the car, I drove 1,107 km yesterday and used an average of 163 Wh/km. Teslafi arrives at 178 Wh/km over the same distance. I wonder where this difference comes from. 10% would make sense if AC/Heat was counted in, so Tesla under-report consumption by reducing it to drive consumption only? Wouldn't make sense to count charging losses either (and would be smaller than that as I've been supercharging only on this trip).
* yeah, it still is a far cry from where it should/could be. One thing I don't get is why Tesla don't harness the power of the masses to improve on this speed sign recognition clusterduck by letting users do simple actions. On an opt-in basis, I think there should be an option popping up on the screen every time the car determines a change to the speed limit should be made, allowing the user to either do nothing (and accept the change), or have the option to undo to last limit (wrong speed detected). Perhaps an additional option "reset to speed from database".
On a different but related note, just drove Brisbane to Sydney yesterday on a roundabout detour route with 2023.26.8 and noticed a few interesting things:
Range prediction has been surprisingly spot on. That's definitely been getting better for the last several releases. Predicted arrival for a 350km away SC was at 1%. I wanted a little more buffer so I planted myself behind a B-double for about 100km and gained 5% for an arrival estimate of 6%. Arrived with 5% at the SC. This is all the more surprising as the last 250km of that drive were on the Grafton - Glen Innes - Tamworth route, a route I've never driven before and that has some massive ascents and descents (by Aussie standards). So Tesla would have no data of mine on that route. In fact on this entire ~2,700 km to Brisbane and surrounds the car overall has provided really accurate SOC predictions. As long as I let AP drive of course. This all goes (ludicrously) rapidly out the window as soon as I drive with play factor.
Speed limit sign recognition worked really well*, except for where it didn't: Had another two random instances of the car reading non-existent 50 signs in a 110. Even saved the video to go back and try and find which sign it misinterpreted. Turns out there weren't any signs visible at all. I can only assume that comes out of an overly imaginative database. Also encountered several really effing annoying repeat instances of the car imposing a speed limit of 80 when it was a sign "80 for trucks and buses". They still don't have compound signs sorted apparently.
Autosteer performed nearly flawlessly. Still swerves on country roads that widen for a turn-off, but really did well at all other times, even in situations where I was borderline confused myself on where to go it managed just fine. Brisbane has a few "interesting" locations with complex road layouts and turning options. AP and I saw those for the first time, and I can honestly say AP drove those better than me seeing them for the first time.
I don't often pay attention to the trip meter data, but on this trip I noticed a discrepancy between what Teslafi and the car report. According to the car, I drove 1,107 km yesterday and used an average of 163 Wh/km. Teslafi arrives at 178 Wh/km over the same distance. I wonder where this difference comes from. 10% would make sense if AC/Heat was counted in, so Tesla under-report consumption by reducing it to drive consumption only? Wouldn't make sense to count charging losses either (and would be smaller than that as I've been supercharging only on this trip).
* yeah, it still is a far cry from where it should/could be. One thing I don't get is why Tesla don't harness the power of the masses to improve on this speed sign recognition clusterduck by letting users do simple actions. On an opt-in basis, I think there should be an option popping up on the screen every time the car determines a change to the speed limit should be made, allowing the user to either do nothing (and accept the change), or have the option to undo to last limit (wrong speed detected). Perhaps an additional option "reset to speed from database".
Talking of improvements, my car recognised a number of overhead LED based speed signs yesterday. This was in Sydney on the M1, M2, Eastern Distributor et al.
Previously my car has decoded none of these but yesterday they were all correct. Some of them could have been geo based but there were several where the speed limit had been dropped because of problems. However, I cannot say these specific signs never worked before because my other experiences were on different roads.
Plus, as we know, there are multiple sign providers with different PWM schemes and flicker rates. But it seems to be another step forward.
Previously my car has decoded none of these but yesterday they were all correct. Some of them could have been geo based but there were several where the speed limit had been dropped because of problems. However, I cannot say these specific signs never worked before because my other experiences were on different roads.
Plus, as we know, there are multiple sign providers with different PWM schemes and flicker rates. But it seems to be another step forward.
Someone on FakeBook mentioned that their TACC/AP exhibited problems . Whether this is directly related to the update remains to be seen
I'm currently on 2023.20.9 and it's appears to be working fine. Read speed signs correctly, even led ones. No blank screen experienced so far, touch wood.
Apparently the AP/TACC became unavailable after 26.8. Now that does not mean they are related.
As with FB posts they are very hard to find once so unable to screenshot the post.
Same version here. So seems like progress on the LED signs (which is definitely not an easy problem to solve).
OzTzlaBloke
Member
Yes I got up to speed on the history there brother. Definite possible maybe (lol as the internet rumour goes plus if someone did know for sure Im pretty sure they would pretend to not know beyond rumour cos reasons right
) is that deep rain is all over as a concept as were post release 4 now on the wipers and its not "actually smart wipers" cos deep rain didnt work we all agree its kinda ridiculous and below what we expect of our beloved brand. Theyve moved onto the approaches they use in all the other training models they use for other systems in AI decision making. That is the training for the neural net is composted all round video training not one feed from one sensor. One sensor which annoyingly gets ditry and then the wipers go spaz mode on ludicrous speed rubbing away on a high frictiondry bit of glass....This is the last system still on their old skool training approach. I get people are over it and I agree. It needs to be fixed and shortly.Similar threads
