Got an update this morning. Just bug fixes and improvements. Is this a new naming convention for updates?
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Software Update 2017.28 C528869
- Thread starter Tozla
- Start date
-
- Tags
- Australia New Zealand
Shock-On-T
⚡️⚡️⚡️⚡️⚡️
According to an Electrek article this morning it's apparently quite a major improvement for AP2 cars.
Tesla pushes new Autopilot 2.0 update with truly ‘silky smooth’ control algorithm
The new naming structure refers to the internal release code, which they will probably use from now on.
I haven't driven it yet because I waited until I got to work to install the update but I will let you know if the "silky smooth" is really silky smooth as I have a few roads where the autosteer is all over the place.
I'll update you all over the weekend.
I went for a drive at lunchtime and I thought that at times it handled staying in the lane much better and then towards the end of my drive it was all over the shop. I thought acceleration and breaking was back to super quick and sudden. I felt this had improved in the last update. For me this is a step back. Not silky smooth at all in my book. Disappointing.
Every big update seems to take a while to settle down- not sure how much is learning or re-calibration but in general I just use AP as much as i can in the beginning to see how it acts after 3-400ks.
With ...76 I found a couple of times it was trying to sit ON the lane markers (slow traffic), but did some previously difficult corners much more comfortably- still letting it bed-in further before i judge it!
With ...76 I found a couple of times it was trying to sit ON the lane markers (slow traffic), but did some previously difficult corners much more comfortably- still letting it bed-in further before i judge it!
Jays200
Member
I got this 2 days ago on my week old MX so don't have anything to compare it to.
But, and a big but, yesterday from lunchish onwards, autopiliot started hugging the right hand line on the freeway and other B roads to the point the car knew it was driving over the line and did nothing about.
I usually stay a bit left in the lane and autopiliot tries to centre but this is now practically unusable as it veers to the right hand line/kerb. TACC braking approaching a stopped car is a bit iffy too.
In the update prior to this new update the TACC breaking was the smoothest I had experienced since getting the car. Today I was almost on top of the stopped car before it broke hard.
I'm new to the world of Tesla and didn't realise that the software needed to be "broken in".
I'm new to the world of Tesla and didn't realise that the software needed to be "broken in".
Shock-On-T
⚡️⚡️⚡️⚡️⚡️
Just FYI, TACC technically can't detect a stopped car ahead of you.
So if the traffic slows down and stops, TACC will stop, but if you come round a corner and everyone is already stopped there are cases where TACC merrily drove into the back of them. Even happened during a Tesla test-drive once apparently.
Shock-On-T
⚡️⚡️⚡️⚡️⚡️
vrykolas
Member
I am really starting to agree with this point, and it should be emphasized more to all. An "adjustment" period is definitely seeming real.
I've noticed this the last few updates. And I'll get a good chance to test AP2 HW on a 3000 mile trip from Dallas to Philadelphia and back (in 6 days)
Last night was even worse than during the day. I had quite a few phantom sudden slow downs as if there was a car directly in front when there was only traffic coming in the opposite direction. At times the car was tightly hugging the right hand side of the lane (RHD Car) and sometimes would veer over the line. Thankfully no cars were coming in the opposite direction at the time.
If new software does indeed need an adjustment period I will persist but it can be very dangerous and I wont use it with the kids in the car.
If new software does indeed need an adjustment period I will persist but it can be very dangerous and I wont use it with the kids in the car.
Jays200
Member
That is the exact same behavior mine is displaying. Didn't start that way, It just had a step change around lunch time and I haven't driven it for the last two days. The wife hasn't used AP in that time to see if it has indeed "learnt". Friend got his update yesterday and is happy with it, so far
mrpseudonym
Member
+1 this latest update gets too close to the RH lane markings. In the Eastern Distributor (Sydney) and on the highway at crows nest it crossed over the RH lane marking completely. I still can't use autopilot from nth Sydney to st Leonards which is becoming the benchmark test track for me for each new release. Even with a car in font showing the way, I have to abort when on the highway crossing falcon st. Note I have to abort.. it would seem to be quite happy to send me into oncoming traffic.
Jays200
Member
Please call Tesla service, as I have just finished doing, as they have raised a case for me and said the technicians are very interested in any step changes after a software update. The more vehicles that exhibit the same behaviour the more it adds to the weight of evidence. Thanks
Jayson
Jays200
Member
Well I thought it was fixed this morning. Back to normal behaviour but then around mid morning back to right hand line hugging and even happily over the line in to oncoming traffic lane
I've noticed that it is fine on some roads and absolutely horrific on others.
In addition sometimes it slowdowns nice and steady as it's coming up on cars already stopped at the lights and other times it comes in very hot. No rhyme or reason to it either. Same goes for acceleration. Some times it's nice and easy other times it zooms off as if it's in a drag race. No consistency at all.
In addition sometimes it slowdowns nice and steady as it's coming up on cars already stopped at the lights and other times it comes in very hot. No rhyme or reason to it either. Same goes for acceleration. Some times it's nice and easy other times it zooms off as if it's in a drag race. No consistency at all.
Similar threads
