-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Poll: Has AP2 18.10.4 surpassed AP1 in terms of performance?
- Thread starter ab26
- Start date
Be nice to get people that currently have both cars. Memory is a tricky thing. 
Ask any psychologist.
Ask any psychologist.IMHO, AP2 passed AP1 with 2017.50– this is when AP2 started being able to follow cars into tight cloverleaf on/off ramps, and AP1 could and still cannot.
BTW I still have both AP1 and AP2. So tempted now to trade in AP1 for AP2.5.
BTW I still have both AP1 and AP2. So tempted now to trade in AP1 for AP2.5.
jimmy_d
Deep Learning Dork
Like @GoTslaGo, for me it was at parity before last holiday season.
My primary need for it is for driving back and forth between LA and SF and my general metric is how much it reduces the energy needed for that drive. When I swapped from AP1 to AP2 last September it seems like the effort ticked up and the number of forced disengagements rose quite a bit. Although because it failed in different places on AP2 than it did on AP1 even though it felt 'worse' it was actually better in enough situations that for me there was more that was 'different' than there was that was 'bad' on AP2. It took time to adjust to the change and by that time the software had improved a lot.
Around the time that 42/44/48 came out I was feeling like AP2 was overall better and the number of forced disengagements between LA and SF was about even, and then 50 came out and AP2 was suddenly better feeling and was taking a lot fewer forced disengagements than AP1 did the previous September. Now with 10.4 it's not even a contest. AP2 on 10.4 is in a different league than AP1 from last September.
I have a friend with an AP1 car that I ride in occasionally but I don't really consider that to be AP1 experience so my AP1 knowledge is about 6 months old now. He doesn't seem to think AP1 has change much recently but I don't know if I trust that opinion. For me AP1 improved a lot during the first 9 months of 2017 so I expect it's even better today.
Before 10.4 it felt to me like AP1 and AP2 were similar animals but with different weaknesses. After 10.4 AP2 no longer feels like it's in the same category. I've only had it for two days but I have a ton of miles on AP1 and on AP2 and I feel like I know it pretty well. 10.4 is quite a big change.
So an interesting side question - for people who still have AP1 and who drive it a lot - how much better did AP1 get on 42 and 44 and 50? Is it still improving by leaps and bounds and if so how much better does it feel today than it did six months ago?
My primary need for it is for driving back and forth between LA and SF and my general metric is how much it reduces the energy needed for that drive. When I swapped from AP1 to AP2 last September it seems like the effort ticked up and the number of forced disengagements rose quite a bit. Although because it failed in different places on AP2 than it did on AP1 even though it felt 'worse' it was actually better in enough situations that for me there was more that was 'different' than there was that was 'bad' on AP2. It took time to adjust to the change and by that time the software had improved a lot.
Around the time that 42/44/48 came out I was feeling like AP2 was overall better and the number of forced disengagements between LA and SF was about even, and then 50 came out and AP2 was suddenly better feeling and was taking a lot fewer forced disengagements than AP1 did the previous September. Now with 10.4 it's not even a contest. AP2 on 10.4 is in a different league than AP1 from last September.
I have a friend with an AP1 car that I ride in occasionally but I don't really consider that to be AP1 experience so my AP1 knowledge is about 6 months old now. He doesn't seem to think AP1 has change much recently but I don't know if I trust that opinion. For me AP1 improved a lot during the first 9 months of 2017 so I expect it's even better today.
Before 10.4 it felt to me like AP1 and AP2 were similar animals but with different weaknesses. After 10.4 AP2 no longer feels like it's in the same category. I've only had it for two days but I have a ton of miles on AP1 and on AP2 and I feel like I know it pretty well. 10.4 is quite a big change.
So an interesting side question - for people who still have AP1 and who drive it a lot - how much better did AP1 get on 42 and 44 and 50? Is it still improving by leaps and bounds and if so how much better does it feel today than it did six months ago?
kingjamez
Member
My AP1 experience only dates back 6 months, but I haven't noticed any changes in performance at all. That's fine by me though, AP1 works well enough for me.
-Jim
jimmy_d
Deep Learning Dork
I was really happy with it (AP1) when I upgraded. Not that it was perfect or that I wasn't hoping for better, but the difference between having it and having to do a long drive without it is just enormous. I bought my first S (shown in my profile pic to the left there) just to get AP and when I sold it I honestly felt that the purchase was totally worth it even if I did end up swapping after only 2 years for the promise of an even better system.
In complete agreement with @jimmy_d
Out of curiousity I looked at my old emails to Tesla ([email protected]) regarding AP1 (Wife and I have been using AP1 since Jan 2016)
2016: Emails were almost all about line/lane recognition failures, and subsequent unusual behaviors. There was one where my wife merged onto the HW, but AP1 would only go the local speed limit (I presume map related issue in retrospect). My recollection is that AP1 had trouble tracking many of the weaker highway lines (white dots on a concrete[white] highway). There were many "take over steering alerts" at this time. Wife ultimately opted to drive primarily with TACC. I would still use both when I had the chance. Toward the end of 2016 most of my emails to Tesla involved inappropriate "red car" warnings.
2017: There were much less emails. I started using AP1 full time, starting pretty much with 17.11.10 (April 2017). Main issue I reported to Tesla was the oft mentioned deceleration under bridge issue.
I do recall there was still lane recognition issues ("take over steering") up until I think 2017.28. That's when I remember the car started to not ask me to take over, and instead kept driving between grey lines. Kinda recall being freaked out a little about that, but impressed that it no longer forced me to take over. I do recall that the software transition did have gradually improved lane recognition, and better maintenance of AP1 on the highway over time... it would stop diving for off-ramps.
I think these iterations also made AP1 smoother in lane, since I did find an email after 2017.42 where I commented that the in lane bouncing had returned. So in retrospect it was definitely better.
with 2017.50, that's when I discovered that AP1 could do gentle off-ramps, after testing AP2 and AP1 on a tight off-ramp (exit angle around 45 degrees). AP2 could handle the initiation of the offramp, whereas AP1 could not. But AP1 could take a shallow offramp (~10 degree). The main thing is that both would adjust their speeds appropriately for the off-ramp. I still remember taking a shallow off-ramp in Jan 2016 where I had to manually take over to get into the off-ramp, and TACC didn't adjust the speed (no car in front), so I almost flew off the off-ramp at 70 mph. I learned to disengage AP going into off-ramps after that.
Ever since 2017.50, I've noticed continued improvement with AP2, with minimal to no further improvement with AP1.
Out of curiousity I looked at my old emails to Tesla ([email protected]) regarding AP1 (Wife and I have been using AP1 since Jan 2016)
2016: Emails were almost all about line/lane recognition failures, and subsequent unusual behaviors. There was one where my wife merged onto the HW, but AP1 would only go the local speed limit (I presume map related issue in retrospect). My recollection is that AP1 had trouble tracking many of the weaker highway lines (white dots on a concrete[white] highway). There were many "take over steering alerts" at this time. Wife ultimately opted to drive primarily with TACC. I would still use both when I had the chance. Toward the end of 2016 most of my emails to Tesla involved inappropriate "red car" warnings.
2017: There were much less emails. I started using AP1 full time, starting pretty much with 17.11.10 (April 2017). Main issue I reported to Tesla was the oft mentioned deceleration under bridge issue.
I do recall there was still lane recognition issues ("take over steering") up until I think 2017.28. That's when I remember the car started to not ask me to take over, and instead kept driving between grey lines. Kinda recall being freaked out a little about that, but impressed that it no longer forced me to take over. I do recall that the software transition did have gradually improved lane recognition, and better maintenance of AP1 on the highway over time... it would stop diving for off-ramps.
I think these iterations also made AP1 smoother in lane, since I did find an email after 2017.42 where I commented that the in lane bouncing had returned. So in retrospect it was definitely better.
with 2017.50, that's when I discovered that AP1 could do gentle off-ramps, after testing AP2 and AP1 on a tight off-ramp (exit angle around 45 degrees). AP2 could handle the initiation of the offramp, whereas AP1 could not. But AP1 could take a shallow offramp (~10 degree). The main thing is that both would adjust their speeds appropriately for the off-ramp. I still remember taking a shallow off-ramp in Jan 2016 where I had to manually take over to get into the off-ramp, and TACC didn't adjust the speed (no car in front), so I almost flew off the off-ramp at 70 mph. I learned to disengage AP going into off-ramps after that.
Ever since 2017.50, I've noticed continued improvement with AP2, with minimal to no further improvement with AP1.
Memory is the worst thing to use when making comparisons.
Here's a video of AP1 vs. AP2 in the same road,same speed, same weather.
This is non-bias, independently verifiable evidence and AP1 clearly wins and this was just 7 days ago.
Here's Part 2 with same time of day
Here's a video of AP1 vs. AP2 in the same road,same speed, same weather.
This is non-bias, independently verifiable evidence and AP1 clearly wins and this was just 7 days ago.
Here's Part 2 with same time of day
Last edited:
suwaneedad
Member
The discussion of AP1 vs AP2 confuses me when I try to read these posts through the filter of EAP and FSD.
Is all of the discussion about AP1v2 performance really a discussion of EAPvEAP? (Does FSD exist on any vehicle yet?)
The reason I am asking is that I've configured Model 3 for EAP, but if these leaps forward with "AP" lately apply to FSD, I want to call my DS tmrw morning and see if I can add FSD to my order to avoid the $1k price difference later.
Is all of the discussion about AP1v2 performance really a discussion of EAPvEAP? (Does FSD exist on any vehicle yet?)
The reason I am asking is that I've configured Model 3 for EAP, but if these leaps forward with "AP" lately apply to FSD, I want to call my DS tmrw morning and see if I can add FSD to my order to avoid the $1k price difference later.
kingjamez
Member
As of today. EAP is all there is. FSD offers no additional features. Yet. (Hopefully)
-Jim
Right. In January 2017, Elon famously promised that AP 2.x ‘FSD would differentiate from EAP’ “soon” meaning those of us that bought FSD would get ‘something’ that EAP people did not have. Not level 4 or 5, but a feature. This was the origin of infamous quote ‘three months probably, six months definitely’. We’re still waiting. And that quote has turned into a meme. And Elon has reused it for other promises.
(I might be paraphrasing very slightly....)
(I might be paraphrasing very slightly....)
The AP2 car in the video is not on the firmware this thread is discussing. The 10.4 update brought much more stability compared to 6.1.
⚡️ELECTROMAN⚡️
Village Idiot
Seriously? But like how many times does that guy have to be right? @Bladerskb I would say that’s a major fail if your video isn’t using the latest update stated in the title.
Not using the (latest) version that this thread is specifically about, is the worst thing to use when making comparisons
dhanson865
Well-Known Member
nothing in the current version is FSD. You'll know FSD when you can verbally tell the car a destination and sit with your hands on your lap while it drives per Autopilot - https://www.tesla.com/autopilot
If you can get out of bed and go grab breakfast without being assaulted by blimps with loudspeakers announcing FSD, it hasn't happened and you can save your money.
In reality the choices are
AP1
AP2 (for both cars prior to 18.10.4 and those that were on the AP1/AP2 cusp and didn't pay for EAP?)
EAP2 (started at 18.10.4)
FSD (doesn't exist yet)
So the poll at the top should really be has EAP2 surpassed AP1? Because there is no EAP1 and AP2 now is enabling enough capability to claim it's first EAP feature.
what we have with 18.10.4 is EAP2 because Autosteer+ is finally enabled (or maybe just finally good enough to be reasonably certain it's enabled), prior to that we had AP2.
Isn't there a handful of cars somewhere with AP2 hardware but no EAP paid for? Test one of those on the same road as a EAP car and that'd be your proof that EAP features are enabled.
I didn't make the connection first. Someone on reddit linked to the EAP page and mentioned the feature on one of the video threads. It matches with the new capabilities we are seeing. In that thread or another I read shortly after somone mentioned the old AP1 and AP2 wouldn't turn the steering wheel past about 20% left or right. The new EAP2 is now turning as much as 60% left or right to handle tighter curves.
the page the image and text comes from is Autopilot - https://www.tesla.com/autopilot
The remaining missing EAP functions are
* Smart Summon
and depending on your opinion if this is here or not
* On-ramp to Off-ramp
I'd say both of those are missing still. Because On-Ramp to Off-Ramp while being under EAP sounds like FSD lite:
Last edited:
⚡️ELECTROMAN⚡️
Village Idiot
⚡️ELECTROMAN⚡️
Village Idiot
nothing in the current version is FSD. You'll know FSD when you can verbally tell the car a destination and sit with your hands on your lap while it drives per Autopilot - https://www.tesla.com/autopilot
If you can get out of bed and go grab breakfast without being assaulted by blimps with loudspeakers announcing FSD, it hasn't happened and you can save your money.
In reality the choices are
AP1
AP2 (for both cars prior to 18.10.4 and those that were on the AP1/AP2 cusp and didn't pay for EAP?)
EAP2 (started at 18.10.4)
FSD (doesn't exist yet)
So the poll at the top should really be has EAP2 surpassed AP1? Because there is no EAP1 and AP2 now is enabling enough capability to claim it's first EAP feature.
what we have with 18.10.4 is EAP2 because Autosteer+ is finally enabled, prior to that we had AP2.
I didn't make the connection first. Someone on reddit linked to the EAP page and mentioned the feature on one of the video threads. It matches with the new capabilities we are seeing. In that thread or another I read shortly after somone mentioned the old AP1 and AP2 wouldn't turn the steering wheel past about 20% left or right. The new EAP2 is now turning as much as 60% left or right to handle tighter curves.
the page the image and text comes from is Autopilot - https://www.tesla.com/autopilot
The remaining missing EAP functions are
* Smart Summon
and depending on your opinion if this is here or not
* On-ramp to Off-ramp
I'd say both of those are missing still. Because On-Ramp to Off-Ramp while being under EAP sounds like FSD lite:
You are missing several features thats supposed to come under EAP
- automatically change lanes without requiring driver input
- transition from one freeway to another
- exit the freeway when your destination is near
- self-park when near a parking spot
- Autosteer+
- Smart Summon
Also being able to marginally handle tighter and curvier road isn't autosteer+. The description reads "your Tesla will navigate tighter, more complex roads." curvy, cresty, dip/hilly roads are not complex roads. Marginal improvement on curvy roads vs AP1 isn't autosteer+
Similar threads
- Poll
