Navigate on Autopilot is Useless (2018.42.3)

[avoigt]

Respectfully disagree. This is not frustration, just facts.

Fact: A double-digit percentage of owners are on AP1.
Fact: AP1 is a benchmark for Tesla's progress on future versions.
Fact: AP1, to-date, has nearly every usable functionality that AP2+ has, plus others AP2+ does not. (AP1+: Autosteer, Auto lane change, blind spot detection, rain sensing, auto high beams, AEB, SCA)

Just because Tesla bumps the version on things doesn't make the previous version irrelevant.

Perhaps if AP2+ could withstand comparison to AP1 with any actual differentiation/improvement, this discussion wouldn't be nearly as relevant.

Respectfully disagree as well

Just Facts

Fact 1: AP1 is not sold any more
Fact 2. Comparison between the two system its not valid as different SW and partly sensors, data a.o.
Fact 3: Most drivers using AP today never experiences or used AP1 so why compare
Fact 4: AP1 will not come back its outdated
Fact 5: I never said AP1 is irrelevant
Fact 6: More AP2x are sold today in a given timeframe than AP1 ever sold in the same timeframe

The previous version is just an outdated one and thats another fact.

So to be clear I do not call AP2x better or worse AP1 but the entire conversation does not add any value for people who want to use an AP or decide not buy one or not. On top of that there are many videos out there I posted a few from users of NoAP that are obviously very happy with the B status which contradicts to that what you wrote.

To add to that you listed a variety of misses from NoAP that I called out never to be claimed from Tesla before.

For some reason you did not reply to that.

 

[GlmnAlyAirCar]

Respectfully disagree as well

Just Facts

Fact 1: AP1 is not sold any more
Fact 2. Comparison between the two system its not valid as different SW and partly sensors, data a.o.
Fact 3: Most drivers using AP today never experiences or used AP1 so why compare
Fact 4: AP1 will not come back its outdated
Fact 5: I never said AP1 is irrelevant
Fact 6: More AP2x are sold today in a given timeframe than AP1 ever sold in the same timeframe

The previous version is just an outdated one and thats another fact.

So to be clear I do not call AP2x better or worse AP1 but the entire conversation does not add any value for people who want to use an AP or decide not buy one or not. On top of that there are many videos out there I posted a few from users of NoAP that are obviously very happy with the B status which contradicts to that what you wrote.

To add to that you listed a variety of misses from NoAP that I called out never to be claimed from Tesla before.

For some reason you did not reply to that.

Regardless of your opinion, AP1 IS the de-facto standard against which AP2, 2.5, etc. are compared. Nearly every post regarding autopilot mentions AP1 as a baseline.

 

[wk057]

Comparison to AP1 adds significant value. It shows people where Tesla is compared to it's previous offerings. (Spoiler: a little over 2 years behind.)

It also gives Tesla a kick in the ass to improve, since they haven't even gotten out of the hole compared to AP1 yet.

AP1 isn't going anywhere. Get over it.

 

[Johan]

Hmm... seeing many reports here om TMC today of people having all AP feature not working, both on Model S and Xs, possibly 3s as well? Tesla doing a fleet wide upgrade and turning of certain HW/SW combos preemtively?

 

[verygreen]

nah, it's just a glitch in one of the telemetry triggers or some other such thingie I am sure. All cars that don't get them that I know of are 100% fine.

 

[S4WRXTTCS]

Did you read the agreement before turning on NoA? It explicitly says to check your surroundings and blind spots prior to confirming. I read this to mean if it is not safe to change right then, do NOT confirm the lane change.

View attachment 350234

The release notes, and the actual behavior of NoA seem to be a bit contradictory.

It routinely asks me permission to change lanes when it's showing red so it knows I can't right then. So why is it asking?

So I have two options

Option A -> confirm the lane change based on the assumption that it will make the lane change when it's safe to do so. With this option I'd oversee it to make sure it's safe.

Option B -> Ignore the request, and indicate a lane change request myself when safe to do so. Assuming of course I want the lane change.

So which option is the right option?

We're in this messy middle where the machine has to ask permission, and the response will likely be delayed by a couple seconds. So whatever was valid back then is no longer valid.

I haven't gone with option A so I can't judge how it does, but it does seem like some drivers are going with this option.

 

[Electroman]

To claim AP2 is not better than AP1 is neither factual nor truthful.

AP1 isn't going anywhere. Get over it.

You got that right. AP2 on the other hand will go places

 

[S4WRXTTCS]

Comparison to AP1 adds significant value. It shows people where Tesla is compared to it's previous offerings. (Spoiler: a little over 2 years behind.)

It also gives Tesla a kick in the ass to improve, since they haven't even gotten out of the hole compared to AP1 yet.

AP1 isn't going anywhere. Get over it.

We have to make fair comparisons with AP1 because AP2 in a lot of ways is what it is DUE to AP1.

For me what AP1 really excelled at was not having any major false positive while TACC only driving. Sure occasionally I'd have minor ghost braking, but nothing serious. With AP2+ there are a significantly more reports of it whether it's AP driving or TACC driving.

Now at first you might say "That's cause AP2 sucks", but that's not really true. It's likely because AP1 drivers kept crashing into things so Tesla made AP2 way too sensitive. Pretty much all the "crashed into a firetruck" are AP1 vehicles.

So what would you give the win to? AP1 which gets Tesla bad press? Or AP2 which can be REALLY annoying? To the point where some drivers even acknowledge they don't use AP with their spouse in the car because they get yelled at any time the car suddenly does phantom braking.

As to automatic lane changes it's hard to compare the two simply because AP1 has no real side/blind spot monitoring. So you're essentially comparing a blind system with a system with eyes or at least 1.5 eyes.

This is likely why lane changes sometimes fail with AP2.

My most recent example of a failed automatic lane change was caused by a combination of AP2 giving too long to indicate the lane change, and the car I was changing in front of accelerating. So I marked it as pass instead of a failure. As in it worked as programmed even though I felt like it was being way too driving miss daisy with lane changes.

It would have been a successful lane change with AP1 though for two reasons. AP1 would accelerate (at least to my recollection), and didn't have cameras to even know what the other guy did.

Sadly I didn't have much experience with AP on AP1 due to the truck lust issues. There were lots of user reports that it got better in V8.X (somewhere in the V8), but by then I was a TACC forever advocate. Now I have a Model 3 so I'm slowly seeing how AP2. 5 does in both AP and TACC.

 

[bhzmark]

AP1 is, in mine and many others' options, still the best publicly available drive assist system out there. AP2 has yet to match its features and reliability fully. This is objectively true, despite claims to the contrary. It doesn't matter if AP2+ can do new-feature-X (poorly) that AP1 doesn't have when AP1 has features Y,Z,A,B,C,D,E that AP2+ hasn't even gotten close to implementing yet.

I own and have driven both quite a bit and ap2 is clearly better at going over hills and allowing cars in front to merge than ap1

 

[Kristoffer Helle]

Tiles is destroying ap1 in Norway.. Useless in many parts of Norway. Keep braking because of wrong speed limits in maps. Getting speed limits from side roads.
Terrible...

 

[Bladerskb]

We have to make fair comparisons with AP1 because AP2 in a lot of ways is what it is DUE to AP1.

For me what AP1 really excelled at was not having any major false positive while TACC only driving. Sure occasionally I'd have minor ghost braking, but nothing serious. With AP2+ there are a significantly more reports of it whether it's AP driving or TACC driving.

Now at first you might say "That's cause AP2 sucks", but that's not really true. It's likely because AP1 drivers kept crashing into things so Tesla made AP2 way too sensitive. Pretty much all the "crashed into a firetruck" are AP1 vehicles.

So what would you give the win to? AP1 which gets Tesla bad press? Or AP2 which can be REALLY annoying? To the point where some drivers even acknowledge they don't use AP with their spouse in the car because they get yelled at any time the car suddenly does phantom braking.

As to automatic lane changes it's hard to compare the two simply because AP1 has no real side/blind spot monitoring. So you're essentially comparing a blind system with a system with eyes or at least 1.5 eyes.

This is likely why lane changes sometimes fail with AP2.

My most recent example of a failed automatic lane change was caused by a combination of AP2 giving too long to indicate the lane change, and the car I was changing in front of accelerating. So I marked it as pass instead of a failure. As in it worked as programmed even though I felt like it was being way too driving miss daisy with lane changes.

It would have been a successful lane change with AP1 though for two reasons. AP1 would accelerate (at least to my recollection), and didn't have cameras to even know what the other guy did.

Sadly I didn't have much experience with AP on AP1 due to the truck lust issues. There were lots of user reports that it got better in V8.X (somewhere in the V8), but by then I was a TACC forever advocate. Now I have a Model 3 so I'm slowly seeing how AP2. 5 does in both AP and TACC.

Wasnt the guy whose AP drove him into a divider using AP2?

 

[Mayhem]

If it works for you, great.

Honestly, I don't think any of these companies would work with me on this, including Tesla (who has offered me a position in the past, surprisingly, but I refused to move to California) due to my stance on not relocating. Making my own company for this.... meh, maybe.

Do you have an avenue for feedback to the Autopilot team? Seems like your insights would be greatly valuable to them. Agree, feels like 'alpha' software. Surprised there was this big of a rollout, but I do like the way the features are heading.

 

[Alphacrux]

I tried v9 autopilot for the first time today. It seemed to consistently want to drive in the right side of the lane, so much that I was uncomfortable using it and turned it off. Has anyone else noticed this? Or, is there some chance that my cameras are misaligned?

 

[Electroman]

Every video I have seen on YouTube has been quite positive and encouraging. It is usually the other way, where internet trolls try to use it in locations where it will have the most difficulty and give a click bait title , "Autopilot tried to kill me".

 

[DarkMatter]

The last time I had an AP1 loaner I accidentally came up with a list of four locations where I regularly use AP2 that were total failures for AP1. One of them was even a "red hands" location. If AP1 is the benchmark for AP2, then I believe the benchmark has been met. The only things I will grant to AP1 still are the quality of the multiple car display (slightly superior for cars in view) and the willingness to stay the course in uncertain situations (which is terrifying in at least one location I can think of). AP2 has started to even close in on proper control in situations that I would never expect any AP1 car to get, such as managing to show accurate lane lines in a bizarre diagonal intersection close to my home. I wish the growth had not taken two years, but now that it is here I am very impressed.

 

[DarkMatter]

I tried v9 autopilot for the first time today. It seemed to consistently want to drive in the right side of the lane, so much that I was uncomfortable using it and turned it off. Has anyone else noticed this? Or, is there some chance that my cameras are misaligned?

That sounds out of calibration. Call your service center and ask them to reset it remotely. If that does not work, schedule a service appointment.

 

[Mark Z]

I made a call to Tesla at 4 AM this morning to complain about Navigation on Autopilot. My points to Tesla:
1. Vehicle slowing on an empty freeway (57 NB Sat AM) because it wanted to move to the HOV lanes was unacceptable.
2. No physical way to disagree with suggested lane change to let the software know that I am satisfied with my lane.
3. Need to have HOV On/Off control button near the Navigation on Autopilot control button rather than drilling down into settings menus IF the "Use HOV Lanes" selection is turned ON.

I look forward to improvements needed for Navigation on Autopilot to make it useful and not frustrating.

 

[azred]

WK057 is a much more "stable genius" than the other more famous one. His posts are great.

I have two Tesla's neither with AP though I have tested it on my S and am on day 2 of a Model 3 trial. I was about to conclude that I needed more time to understand the new NOAP but see it may not be the case.

For some reason I am still tempted to plunk down the $5500. Obviously I am unstable non-genius.

 

[Electroman]

WK057 is a much more "stable genius" than the other more famous one. His posts are great.

I have two Tesla's neither with AP though I have tested it on my S and am on day 2 of a Model 3 trial. I was about to conclude that I needed more time to understand the new NOAP but see it may not be the case.

For some reason I am still tempted to plunk down the $5500. Obviously I am unstable non-genius.

First you need to settle down with regular EAP for a few weeks. Then you can try to NoAP

 

[NuclearPowered]

So with all the love for AP1 and the disdain for AP2, just what exactly is (stock) AP1 better at? I've driven both and AP2 excels in situations in which AP1 would drive off the road or throw up the red hands. NoA has its issues and isn't super useful in the current state, but overall it's not that bad.