So SilkySmooth never happened now waithing on SomethingSpecial

[AnxietyRanger]

At present, if you value your sanity, forget about using autopilot in construction zone. It's crazy enough that ap sees the black line over white road marks as the better part but when road shifts suddenly in construction zone, is an accident waiting to happen While using ap.

Sure.

I think next to the ghost brakings (nothing to do with construction zones), the tendency of AP to grab onto false road markings is the second biggest issue. They can and do of course happen in construction zones, but that is not the only place. Basically anywhere where the road may have been fixed a little crudely AP can dismiss the beautiful white lines and suddenly starts following some traces of tar... There are several of such reports on TMC...

 

[Electroman]

Here's my brief don't do it for AP2
---------------------------------------------
Construction Zones
Very wide single lanes with yellow solid on left and white solid on right
Crossing intersections without a tracking car
Approaching intersections without a tracking car
Exit Ramps
Very curvy local roads

Other then these, I find it to work quite well

And pretty much a weak spot for AP1 also.

 

[Electroman]

Sure.

I think next to the ghost brakings (nothing to do with construction zones), the tendency of AP to grab onto false road markings is the second biggest issue. They can and do of course happen in construction zones, but that is not the only place. Basically anywhere where the road may have been fixed a little crudely AP can dismiss the beautiful white lines and suddenly starts following some traces of tar... There are several of such reports on TMC...

I think this is where AP1 shines. It figures out the lanes even if they are faded quite well. Even bots are not a problem as long as there are bots at regular intervals

 

[mandersen]

I think we need to give up on the rain sensing wipers...

Wait no auto sensing wipers? Seriously? Just ordered and moving from an Infinity Q50S. Living in WA the auto sensing wipers are a needed necessity. I am sure that it spells out it has them in the manual.

 

[AnxietyRanger]

I think this is where AP1 shines. It figures out the lanes even if they are faded quite well. Even bots are not a problem as long as there are bots at regular intervals

Yes, sure. I am talking about AP2 as I assumed the thread is about that.

AP2 tends to - at times - prioritize any dark lines on the road over bright white road markings and that's an issue.

 

[AnxietyRanger]

Wait no auto sensing wipers? Seriously? Just ordered and moving from an Infinity Q50S. Living in WA the auto sensing wipers are a needed necessity. I am sure that it spells out it has them in the manual.

AP2 does not have them, as it apparenly has camera-based rain detection which was supposed to come out in December 2016 (it was listed as one of the features that "will" come in December 2016), but it is not out yet.

On the Autonomous Vehicles sub-forum on TMC the difficulty of this rain sensing is discussed and some people are pessimistic about it.

I'm still optimistic they'll get something done. Whether or not it will be good or more like AP1 blind spot detection "good", we shall see.

 

[bedoig]

Wait no auto sensing wipers? Seriously? Just ordered and moving from an Infinity Q50S. Living in WA the auto sensing wipers are a needed necessity. I am sure that it spells out it has them in the manual.

Just some good-natured ribbing:

Nothing is more necessary than a needed necessity...

No auto-wipers... major first world problem...

I get where you're coming from though

 

[mandersen]

Just some good-natured ribbing:

Nothing is more necessary than a needed necessity...

No auto-wipers... major first world problem...

I get where you're coming from though

LOL! We have 4 cars in our household and I will get so much crap if the one that costs 2x+ all the others doesn't have it.. going to be painful my kids are teenagers and like to poke the bear!

 

[SlicedBr3ad]

Just some good-natured ribbing:

Nothing is more necessary than a needed necessity...

No auto-wipers... major first world problem...

I get where you're coming from though

Couldn't agree more. I got auto rain sensing wipers in Infiniti and I have to say, gimmick. I'm in state Florida so it would have been useful if it worked. instead, its more hassle than it's worth and I find myself using wipers manually more often than not. Those complaining about this feature not being in Tesla ap2 hardware must know something I don't.

 

[Electroman]

It works flawlessly in my 11 year old Lexus..

 

[jddssc121]

Couldn't agree more. I got auto rain sensing wipers in Infiniti and I have to say, gimmick. I'm in state Florida so it would have been useful if it worked. instead, its more hassle than it's worth and I find myself using wipers manually more often than not. Those complaining about this feature not being in Tesla ap2 hardware must know something I don't.

I've had it in all my cars since 2006 or so (i'm embarrassed to say how many cars that is) and it's always been flawless. I really miss it.

 

[Shock-On-T]

We are in the uncanny valley of autonomous vehicles.
Drive attentively for a few more year, and stay safe.
The cars are getting better and better, but right now it's risky relying too heavily on AP.
As an AP1 driver I want to thanks to the AP2 guys for improving my next car!

 

[Mattzilla]

In about 50k miles on AP1/TACC, I've only had 2 obvious ghost braking events. Both were annoying and one almost caused the person behind me to rear end my car. I made sure to send apologetic glances to the other driver afterwards for the incredible anxiety that the sudden slamming on of the brakes caused both of us.

Sounds like it happens much more often on AP.2.

So the obvious solution right now is to stick with AP1, unless you enjoy being a lab rat for AP 0.2 which is obviously still in early development/alpha.

 

[AnxietyRanger]

So...does that mean I should have tesla remove EAP before I take delivery of my new car??? This thread makes me nervous...if it works, it's great, if it doesn't work, you'll get rear ended or worse.

IMO get the EAP if you want it, but just know what you are getting into and use it when you think you can seriously keep an eye and foot on it. I even got FSD, but I don't know how justified that is until some features using the FSD cameras appear.

So the obvious solution right now is to stick with AP1, unless you enjoy being a lab rat for AP 0.2 which is obviously still in early development/alpha.

Yeah, if you are on AP1, do not upgrade to AP2 yet. There is no point. Model 3 is introducing new things to Teslas anyway, so wait patiently at least until Q1/2018 I think before beconsidering an upgrade and see what new goodies make it to Model S/X and how AP2 (or 2.5 or 3...) matures in the meanwhile.

If for some reason one must buy a new Tesla before Q1/2018, I would still go for AP2 instead of AP1 for the future updates that are bound to come (barring a company failure), but mostly I would just wait if I were in the buying game (which happily I am not, buying a Tesla is far too stressful).

 

[Tangible]

I would just wait if I were in the buying game

As one who bought in late 2016, I would add to this good advice:

...and only buy based on features that are already in production and working well. Pay no attention to what Tesla promises or "expects", as they have shown that they should not be trusted.

 

[AnxietyRanger]

As one who bought in late 2016, I would add to this good advice:

...and only buy based on features that are already in production and working well. Pay no attention to what Tesla promises or "expects", as they have shown that they should not be trusted.

While I agree this is overall good advice, the problem is that some value to future promises must still be attributed to be accurate. Paying "no attention" is probably not good. Paying "suitable" attention is better.

For example, based on your advice literally, buying an AP1 CPO (or inventory if such exists still) would make clear sense, because it is superior to AP2 at this stage.

But in reality, the likelihood of AP2 over time (short-term, even) being superior to AP1 is sufficient that it has to be factored in, somehow. How exactly, that is a good question (and depends on individual factors too like how long you are planning on keeping the car), but it should feature in the calculus.

I'm sure you didn't mean things literally, anyway, so not disagreeing with you but kind of adding to the thinking.

 

[daktari]

Wonder if the new direktør of AI, Karpathy, was hired because they have identified the troubles they have with improving the specificity.
Btw, my experience is that Ap1 tracks car in front more often than Ap2 when lanes are poor. And ap2 lane recognition has been going backwards since 17.17.4. Hope the new one gets better.

 

[AnxietyRanger]

Wonder if the new direktør of AI, Karpathy, was hired because they have identified the troubles they have with improving the specificity.
Btw, my experience is that Ap1 tracks car in front more often than Ap2 when lanes are poor. And ap2 lane recognition has been going backwards since 17.17.4. Hope the new one gets better.

It is possible EAP does not tell us much about the real current development / future status of AP2.

FSD is a separate codebase and EAP probably only exists because Tesla could not get MobilEye on board to include it on AP2 boards for legacy support, so that AP1 could have continued in the meanwhile while FSD is being developed. We know MobilEye was originally supposed to be there, until the companies parted ways (over the Brown crash and/or data sharing disagreements).

So sometime between spring/summer 2016 and October 2016 Tesla decided to implement a separate EAP codebase to replace AP1 in the short-term. That is what we have now. It is possible it does not tell us much anything about AP2 performance on the FSD codebase.

Where on the AP2 board does one think MobilEye was supposed to go...?

 

[zambono]

You should try AP2 TACC first before saying that...

I guess, but I have to believe that within maybe another 6 months it will be in much better shape. Elon was definitely unrealistic with their original time estimate but they are getting closer now. Or so it seems based on people on here, I have not tried AP2

 

[R.S]

My current MS has AP1, so I can't add any content yet, but I am planning on getting FSD on the wife's 3 and at least EAP on my next S (will be another lease so probably no FSD unless FSD is better than EAP in about a year)

The real problem for me is that Tesla obviously tries to go incremental, despite that being potentially dangerous.

Other car companies have large test fleets, where payed engineers test the limits of the cars. But I as a Tesla customer am supposed to pay 5/8k for beta testing?

I think it's cool that they can upgrade my car, once it's possible. But please, make sure it really works first!