So, when is APv2 going to be noticeably better than APv1? Or is it already?

[IFought4U]

I am not happy with AP2, either. I purchased my S [and ordered a 3] because I care about AP. With better hardware and software upgrades, I have hope.

I only wish (1) the AP could hold the lane on Sunset drive reliably [which, given how many Teslas are driving within lane on sunset every day, should be easy], and (2) deal well with cars on neighboring lanes that cross into my lane. this is not good even on the freeway, where it should be easy. (at times, the car also has tended to want to lurk suddenly.) .

alas, let's give them some time. it's hard to keep self-driving engineering teams intact at the momen. we hope it will improve.

if there is one aspect worth complaining about, it is the lack of communication from tesla with its owners (other than sales claims, of course). I have no idea of what the current AP version is, nor when they plan to release the next major and minor versions. sort of completely in the dark...

I concur. My biggest frustration is having to learn through trial and error what doesn't work. Tesla tells me nothing. All Tesla reps laugh and giggle and say, it'll get better. Pull back the cloak and tell us what we've got and what we can expect. I'm gonna get hurt out here

 

[scottm]

I agree! Intel change of hands of the AP1 technology gives Tesla a way to save face while getting something back ... It's sitting over in Santa Clara on a bench and maybe by now Intel has put enough of their own juice into the tech it could possibly pass as "Intel Inside" instead of EyeQ ..

Or, in a few weeks AMD will have cloned it, so Tesla could buy the vision tech from them for a few bucks less per car.

I called it !!!

Report: Tesla is Working With AMD on New Chip Designed for Autopilot

 

[SkiTrak]

That is actually not necessarily a problem for AP2. Tesla designed AP 2 to have the processor board swapped if it is necessary to fulfil the needs of FSD. Elon has been quoted as saying they would if necessary. I think the people who paid for FSD will get it since they paid for it. Those that decided to wait will probably have to pay more then the undiscounted MSRP Tesla originally set.

As for Tesla abandoning the early adopters that will most likely not make sense. Between the Class Action law suits which make Law Firms rich and the bad PR it would give Tesla it will probably make more sense to spend the money on upgrading the purchasers then defending the Law Suits and repairing their image not to mention what the settlements will end up.

Remember a large part of the cost of FSD will be the software that does not really cost more to put on more cars other then increase support load. I suspect the actual cost of upgrading some sensors and adding a couple of new ones will not be that bad.

 

[dknisely]

That is actually not necessarily a problem for AP2. Tesla designed AP 2 to have the processor board swapped if it is necessary to fulfil the needs of FSD. Elon has been quoted as saying they would if necessary. I think the people who paid for FSD will get it since they paid for it. Those that decided to wait will probably have to pay more then the undiscounted MSRP Tesla originally set.

As for Tesla abandoning the early adopters that will most likely not make sense. Between the Class Action law suits which make Law Firms rich and the bad PR it would give Tesla it will probably make more sense to spend the money on upgrading the purchasers then defending the Law Suits and repairing their image not to mention what the settlements will end up.

Remember a large part of the cost of FSD will be the software that does not really cost more to put on more cars other then increase support load. I suspect the actual cost of upgrading some sensors and adding a couple of new ones will not be that bad.

I'm 100% certain the early MS/MX adopters will be abandoned. It makes no business sense to worry about a few 10Ks of water under the bridge when you are running a M3/MY company. Now if M3 keeps shipping with useless HW, they might get something because of volume. They will settle the lawsuits with a few coupons good for Tesla store merchandise and move on. Most of the chumps will turn around and buy a newer Tesla anyway (possibly to be fooled all over again). Tesla is immune to bad PR. All the PR channels (youtubers, bloggers, etc.) will just get free Teslas and continued access to Tesla events, which they can't live without. The settlements will be nothing.

"FSD" will never be approved, so they are out zero for that. For EAP, they will declare victory and go home. What's the "value" of the various "EAP" features in a vacuum -- perhaps a few $1Ks. I predict $2500 credit toward future Tesla purchases, non-transferable (and agreement to speak nothing of it forever, obviously).

 

[Brian Anderson]

Exactly. A very calculated decision to defraud "rich" MS/MX customers out of money to give themselves working capital for M3 production. They will deal with the fallout later when M3 is generating cash. They will likely get away with just returning the FSD money, but I hope they have to pay penalties also.

That is giving Tesla's engineering and management teams WAY too much credit. More likely is that they honestly believed their timelines and technical capabilities and simply way underestimated.

 

[Kant.Ing]

One way (or maybe the only way) to make AP2 more "enhanced" than AP1 is to replace AP1's codes with AP2's.

Talking about the AMD engagement and Tesla's own new AI chip, wanna bet we will see AP3 all the way to AP5 in the next few years? Below was my prediction 5 months ago about Level 2~5 milestones.

"AP2 L2: end 2017, L3-5: never.
AP3 L3: end 2018, L4-5: never.
AP4 L4: end 2022, L5: never.
AP5 L5: end 2025, and cars on strike at the beginning of 2026!"

Let's aim for "November 13, 2026."

 

[Richk]

Piling onto this thread. When my wife and I got our first MS (2015 70D AP1), the original Autopilot had just released. And it was pretty shaky in the beginning. But over just a handful of months, the software upgrades made it better and better, through 7.0, 7.1, and 8.0, to the point where it was very reliable in traffic.

We thought AP2 would be an "upgrade" when we got our current 2017 90D. I also understood it would take time to build history with the new hardware. But AP2 has been out for almost a year, and I think there's been one update in that time that's made the experience slightly better, but not much.

When AP2 and EAP was first announced, I was excited and the possibilities of AP2 jumping past highway only driving, even ignoring full autonomy. But none of that has come to pass and we still have dangerous steers, in-lane left-right oscillations, TACC that abruptly slows the car for no reason. And no auto-wipers!

Still love the car for 100 other reasons. But I'm a bit let down to be looking forward to the software update that merely puts AP2 on par with what I had a couple years ago.

What's going on that progress has slowed so much? Are the machine vision and AI devs at Tesla getting pulled off to write code for M3?

 

[xav-]

It doesn't really matter whether you have 1 camera or 8.. when you have bad code you can bring all the hardware you want it won't solve anything. If the Tesla engineers can't even handle 1 camera supporting 8 will likely result into a user experience that is worse.

IMO self driving requires Lidar. You can't do this stuff with cameras/radar only. It's a freaking car, false positives and false negatives are unacceptable. With Lidar you get a full 3D picture of the environment. It doesn't even matter that you recognize what that large object 10 yard in front of you is. You know you have to stop. The software on the backend becomes easy.

This whole radar thing introduces a ton of complexity by forcing to over rely on image recognition to vet things. A radar is an awesome feature as a driving assist. It can do amazing things such as looking at several cars in front of you. But for self driving it's way too limited. I will never ever buy a car without a radar again. TACC is another awesome feature.

So even mobileye in my opinion cannot handle self driving. To that you add the fact that the Tesla engineers are trying to do in 6 months what intel has paid $15 billion for, you know that parity is not anytime soon.

 

[bob_p]

We're approaching one year since the first AP2 cars were delivered.

We still don't have rain-sensing wipers - and it appears newer cars (AP2.5?) may now have the separate rain sensor. Has Tesla made any comments about providing the automatic wipers?

Our car is stuck on release .32 - and I've pretty much given up on using AP2 in any drives. It seems to be more work to keep AP2 from crashing into something (especially when there's been recent or ongoing construction) than any benefit AP2 might provide. And on surface streets, at least in our area - it's worse. Plus, the change to the speed limit database creates major challenges when the freeway is marked incorrectly at a lower speed - and the car immediately slows down - in the middle of traffic.

We have a Model 3 reservation - and have been looking forward to getting one in Q4 this year. But a major reason why my wife wants a Model 3 is to have EAP/FSD for her work commute. Based on our experience with AP2 so far, we're concerned about purchasing another Tesla, with an expectation to have EAP/FSD - and end up with two cars that will fall far short of expectations.

As someone who's managed large software/hardware projects, I can appreciate the complexity of what Tesla is trying to do with EAP/FSD - and the risks in trying to deliver major new software capabilities as quickly as possible. And Tesla won't be the first company to badly miss their software schedules.

But, it's been almost a year - and not only is AP2 not close to meeting the promised goals - Tesla has already started to introduce hardware changes, which introduces increased uncertainty on ever getting AP2 working.

If they aren't close to delivering on their promises for AP2 - then let us know where things stand - and re-set expectations...

 

[dknisely]

One way (or maybe the only way) to make AP2 more "enhanced" than AP1 is to replace AP1's codes with AP2's.

Talking about the AMD engagement and Tesla's own new AI chip, wanna bet we will see AP3 all the way to AP5 in the next few years? Below was my prediction 5 months ago about Level 2~5 milestones.

"AP2 L2: end 2017, L3-5: never.
AP3 L3: end 2018, L4-5: never.
AP4 L4: end 2022, L5: never.
AP5 L5: end 2025, and cars on strike at the beginning of 2026!"

Let's aim for "November 13, 2026."

I don't think they will get AP3 out this year, but otherwise, pretty accurate.

 

[dknisely]

We're approaching one year since the first AP2 cars were delivered.

We still don't have rain-sensing wipers - and it appears newer cars (AP2.5?) may now have the separate rain sensor. Has Tesla made any comments about providing the automatic wipers?

...

Yes, no rain-sensing wipers, and the feature was removed from the website, so it is simply gone. The window of people like me who were promised auto-wipers is small, so they can easily dismiss us (and will). After about January 2017, the feature wasn't promised.

However, there is NO documentation that AP2.5 has the rain sensor back. This is an urban myth. The M3 sensor that was spotted is a temp/humidity sensor that is on AP2 as well. It might be that consumer M3s get the rain sensor back, but we have no evidence so far.

They will probably turn on some kind of crude "turn on the wipers any time (E)AP can't see well just to be safe" method, but it won't be robust enough to market as auto-wipers. I don't care about FSD since that is never going to happen on AP2 or AP2.5 cars.

 

[dknisely]

Piling onto this thread. ...

What's going on that progress has slowed so much? Are the machine vision and AI devs at Tesla getting pulled off to write code for M3?

Probably the AI devs are focusing on FSD for the semi truck. It is absolutely essential, and they can throw as much hardware as needed at it (plus the highway-only geofenced nature of truck routes make the job easier). The routes will all be pre-scanned and updated frequently, and they can even add a capture LIDAR to the truck trivially. Also, they can pay for a lot more data backhaul, etc.

 

[TES 90D]

Is it possible for Tesla to install AP1 software into AP2 cars until they get AP2 sorted? As the AP2 cars have more cameras and a bigger computer, and as I understand it, AP2 cars aren't using the extra cameras yet anyway, then it should be possible.

I would like the option of going back to the cruise control that doesn't have the traffic aware component. It was the best cruise control system on any car I've owned, but TACC keeps propping for no reason, so I rarely use it. Not having cruise control is a major bummer for me.

 

[idleuser]

Is it possible for Tesla to install AP1 software into AP2 cars until they get AP2 sorted? As the AP2 cars have more cameras and a bigger computer, and as I understand it, AP2 cars aren't using the extra cameras yet anyway, then it should be possible.

I would like the option of going back to the cruise control that doesn't have the traffic aware component. It was the best cruise control system on any car I've owned, but TACC keeps propping for no reason, so I rarely use it. Not having cruise control is a major bummer for me.

No, Elon already said it's a waste of time and resource.

Elon Musk on Twitter

 

[lunitiks]

Class Action law suits which make Law Firms rich

What's wrong with rich Law Firms? They're supposed to be rich!

 

[Kant.Ing]

No, Elon already said it's a waste of time and resource.

Elon Musk on Twitter

That nVidia platform burning 250 watt seems a big waste of resource to me!

 

[TaoJones]

What's wrong with rich Law Firms? They're supposed to be rich!

/digression

Show me a law firm that's going broke and I'll show you a law firm with a conscience.

Some of the best lawyers I've known are far from rich. But that's partly because I try to not knowingly associate with complete scumbags.

Now, for our first conundrum, how to qualify government lawyers... clearly, they work as hard as anyone else, but don't get paid jack, relatively speaking. And then there are the pro bono folks, who've invested at least some amount into their law degree and who feel compelled to give back. It's an interesting spectrum to say the least. For some, it depends to a degree upon which side of the room you sit.

Cue the "good" lawyers in 3, 2, 1...

/end digression

So now that the P85+ with AP(1) has become the unicorn d' tutti unicorns, who wants to part with theirs for a motivated AP2 car owner (see sig below for details).

Yes, unicorn d' tutti unicorns.

Anyway, much of the frustration could be dispelled with a simple series of cogent and considerate corporate communications to the affected subsets of owners.

 

[Mark Z]

I was live on the phone with Tesla at 9 PM when AP2 changed lanes without the turn indicator. Tesla pulled logs. Tesla called later to have the vehicle brought in for service. TIP: Remember the date and time of such an occurrence. I was asked for that data by the service advisor. Thankfully, the iPhone had the call data in “Recents” to help the Service Center research the log data from Fremont. I don’t expect anything more than an alignment check of cameras. I did suggest they could install 2.5 hardware if necessary!

 

[Matias]

it appears newer cars (AP2.5?) may now have the separate rain sensor. ..

Can you provide a link/source for this?

 

[Matias]

Is it possible for Tesla to install AP1 software into AP2 cars until they get AP2 sorted? .

AP1 uses MobilEye's proprietary software.