AP2 / 8.1 : "Cruise Not Available"-AP intermittently not working in 17.11.{3,10} versions

[Vger]

OK, so the trip happened, and AP2 remained in play the whole time!! Yay! I finally got to use and assess the state of this next-generation AutoPilot. It is a mixed blessing, still, after months in the field.

While it certainly helped on the divided highway sections of the trip, I find its performance is still lagging noticeably relative to the AP1 MobileEye-based system. It very "polygonal" in its approach to sweeping curves, sometimes reacting really late to the approaching lane line, and then lurching sharply across the lane to correct. One could worry about being pulled over for "weaving" with such behaviour!

It is also annoying that AP2 still limits driving at 50 km/h on anything that it does not consider a highway. This leaves a lot of lesser highways uncovered, unlike the AP1 system.

 

[TheTron]

Maybe we all took delivery around the same time? Jan 3rd for me.

First week of Feb for me.

OK, so the trip happened, and AP2 remained in play the whole time!! Yay! I finally got to use and assess the state of this next-generation AutoPilot. It is a mixed blessing, still, after months in the field.

While it certainly helped on the divided highway sections of the trip, I find its performance is still lagging noticeably relative to the AP1 MobileEye-based system. It very "polygonal" in its approach to sweeping curves, sometimes reacting really late to the approaching lane line, and then lurching sharply across the lane to correct. One could worry about being pulled over for "weaving" with such behaviour!

It is also annoying that AP2 still limits driving at 50 km/h on anything that it does not consider a highway. This leaves a lot of lesser highways uncovered, unlike the AP1 system.

OK, so the trip happened, and AP2 remained in play the whole time!! Yay! I finally got to use and assess the state of this next-generation AutoPilot. It is a mixed blessing, still, after months in the field.

While it certainly helped on the divided highway sections of the trip, I find its performance is still lagging noticeably relative to the AP1 MobileEye-based system. It very "polygonal" in its approach to sweeping curves, sometimes reacting really late to the approaching lane line, and then lurching sharply across the lane to correct. One could worry about being pulled over for "weaving" with such behaviour!

It is also annoying that AP2 still limits driving at 50 km/h on anything that it does not consider a highway. This leaves a lot of lesser highways uncovered, unlike the AP1 system.

Interesting that it seems to have reappeared for you. What firmware version are you on and did you experience the exterior lights being constantly on while you had the issue?

I've noticed the IC booting up a couple of times over the last day or so. I'm trying to get the car to sleep more on the off chance that this improves the odds of AP reappearing when the car wakes. I don't think my car ever slept before because TeslaFI was constantly logging.

Like many I'm doing lots of miles over the Easter weekend, gutted that I can't at least use TACC for all those boring highway miles.

 

[JoeZ]

light always on on auto maybe an indication of sensor failure (most likely software issue, because auto light works great before 8.1) previously someone said in the forum that the logs behind his non-functional AP car show camera always calibrate in "low light mode". My guess is the update tricked the sensor thinks it is dark all the time that's why auto light will always on. I am still calibrating my main camera under 17.14.23 after SC restarted the calibration for me. I can't tell if the calibration will work but the auto light bug still exists under 17.14.23.

 

[zonehawk]

Just got my car back from SC yesterday. I received the 17.43.23 update which did not correct the issue, and the SC did not have a fix for the issue except for telling me to wait for the Software Fix.

Worked fine after 8.1 for a week or so but now approaching 2 weeks without any autopilot features or even basic cruise... my daily drive is about 90% highway too, so it's getting old always holding down the accelerator.

 

[Doubletap67]

Just got my car back from SC yesterday. I received the 17.43.23 update which did not correct the issue, and the SC did not have a fix for the issue except for telling me to wait for the Software Fix.

Worked fine after 8.1 for a week or so but now approaching 2 weeks without any autopilot features or even basic cruise... my daily drive is about 90% highway too, so it's getting old always holding down the accelerator.

Did they attempt camera calibration? Apparently that's why my SC will do Monday.

 

[croman]

OK, so the trip happened, and AP2 remained in play the whole time!! Yay! I finally got to use and assess the state of this next-generation AutoPilot. It is a mixed blessing, still, after months in the field.

While it certainly helped on the divided highway sections of the trip, I find its performance is still lagging noticeably relative to the AP1 MobileEye-based system. It very "polygonal" in its approach to sweeping curves, sometimes reacting really late to the approaching lane line, and then lurching sharply across the lane to correct. One could worry about being pulled over for "weaving" with such behaviour!

It is also annoying that AP2 still limits driving at 50 km/h on anything that it does not consider a highway. This leaves a lot of lesser highways uncovered, unlike the AP1 system.

What's odd is that my AP2 ping pongs like crazy around the highways around Chicago which I take every day and where there are thousands of Teslas. I just drove to Ithaca NY and AP2 was awesome! It didn't ping pong that much on I-90 which makes sense but it also doesn't do that on local highways around here (divided and still getting 80mph AS). I'm not sure why that is. It definitely doesn't understand how to set an angle on the steering wheel and do adjustments as the curve progresses. It, instead, seems to take a curve as 8 different zig zags.

Once my car almost fishtailed when it tried overcorrecting at 80mph. I now only set AS at 75mph. I don't trust it at 80. I also lower the speed to 65 on curves I know it does a poor job mainly because it can be nauseating to ping pong around. I'm basically only using AS sparingly which is unfortunate for its price tag. I'm only on 17.11.3. I'm hoping I'll eventually get a newer FW that addresses this major flaw (and hopefully increases local AS's max speed to 45mph or AP1's +5mph).

 

[miken79]

Dropped mine off at the SC a couple hours ago. They had me take them on a test drive to show them. The entire time there's just this aura of "stop lying about AP not working" and so many comments about how the car just can't see the lines on the road? The technician stopped once I asked how that affected cruise control though. I know they aren't happy with me, the only way I ever get a response is by calling customer service instead of them. So depressing. Anyway, the promised at update by 4, about 2 hours from now.

 

[miken79]

There going to replace the autopilot ECU.

 

[JoeZ]

Dropped mine off at the SC a couple hours ago. They had me take them on a test drive to show them. The entire time there's just this aura of "stop lying about AP not working" and so many comments about how the car just can't see the lines on the road? The technician stopped once I asked how that affected cruise control though. I know they aren't happy with me, the only way I ever get a response is by calling customer service instead of them. So depressing. Anyway, the promised at update by 4, about 2 hours from now.

Me too, they acted like they know nothing about this issues

There going to replace the autopilot ECU.

did your autopilot work before 8.1? If they are going to replace your ECU it should be a hardware issue rather than software issue. Please report back if replacing ECU helped.

 

[chillaban]

Me too, they acted like they know nothing about this issues

did your autopilot work before 8.1? If they are going to replace your ECU it should be a hardware issue rather than software issue. Please report back if replacing ECU helped.

I am curious too. I mean if 8.1 really did activate additional cameras it could be pushing the AP ECU to a new performance level and exposing problems that the last version didn't.

 

[miken79]

Me too, they acted like they know nothing about this issues

did your autopilot work before 8.1? If they are going to replace your ECU it should be a hardware issue rather than software issue. Please report back if replacing ECU helped.

So I ran over to grab my toll pass and stuff and got to talk to the tech for a bit. Apparently Tesla is aware of the issue but only for ~240 cars of which mine wasn't on that list. He talked with the lead on the AP stuff and was told the fix will definitely be out with the next firmware but that replacing the ECU would get me back to working now instead of waiting. They also had to put a shim in for the front camera as mine wasn't lining up properly but that shouldn't have effected this, just another issue to add to the list. They're going to get it all replaced then handle doing the 100 miles of driving to get the cameras calibrated again before I see my car Monday evening. I'll post back as soon as I get a drive in it.

At least this time it felt like the tech had a handle on it, can't say that about the previous discussions I've had.

 

[Ulmo]

Today on 17.11.10, across 6 drives, my AP2 features were working the whole time, as far as I know. I was aware of them frequently during many parts of my drive, so I do not recollect them not working today at all. Previously, they had failed on 17.11.10 and 17.11.3, and the failure covered around 80% to 90% of my drives. I do not know if the bug is still intermittent, or has been cured somehow. I have no idea if a hacker hacked in, broke it, and then figured out how to fix their hack to unbreak it but still maintain control of my car, or if it was some other sort of Tesla bug.

So, ever since that Tuesday Apr 11, my AP2 software has been working (today is Saturday Apr 15): speed limit signs, headlamps, lines for the sides of the road, other cars when it can detect them in front, etc.. Whatever problem I had seems to have gone away on its own. I don't know what was causing it. Maybe someone manually logged in to my car from the mothership and cleaned it up, or it reset itself, or a patch came in, or it fixed itself, or the issue causing it has been hidden, or whatever. But, I can say without anything done on my part the problem has not come back since it went away Tuesday.

Apparently Tesla is aware of the issue but only for ~240 cars of which mine wasn't on that list.

Ok. It seems that apparently they figured out that my car was one of the cars affected (even though I don't recall telling them about it), and they somehow fixed it from their side. This explains why mine suddenly started working at a specific time and has worked since: it appears they figured out how to fix the problem and fixed mine (remotely). This must have been via software sometime between Apr 10 Monday at 11:32PM (which was the last time I drove it when it didn't work right) and Apr 11 Tuesday at 2:02PM (which was the first time I drove it which it worked right again after many days of not working, and has been working since).

I received no firmware update notices on the car whatsoever, so whatever it was must have been quiet.

My vehicle was also in for 12,500 mile service on Apr 6, before they fixed it remotely on my car, during the period in which it wasn't working at all for me; I didn't tell them about it since I knew the bug was widespread and would eventually be fixed for everyone. The car was returned to me in the same state (the bug still present). It didn't get fixed for ~5 days after service was complete.

My car is parked in wifi at home, and usually has access to LTE when on the road.

I wish someone told you what the name of the bug is, so we can call it the "SDJ92D" bug or something, so we can just call them up and say "hey I have the SDJ923D" bug or whatever, rather than trying to give a random description to Service and hoping they know how to match that to their internal information. But, once any of us with the bug figure out how to describe it to service, it seems like that's all it will take for them to put you in queue to have them root out the bug from your car (remotely).

 

[Doubletap67]

So, ever since that Tuesday Apr 11, my AP2 software has been working (today is Saturday Apr 15): speed limit signs, headlamps, lines for the sides of the road, other cars when it can detect them in front, etc.. Whatever problem I had seems to have gone away on its own. I don't know what was causing it. Maybe someone manually logged in to my car from the mothership and cleaned it up, or it reset itself, or a patch came in, or it fixed itself, or the issue causing it has been hidden, or whatever. But, I can say without anything done on my part the problem has not come back since it went away Tuesday.

Ok. It seems that apparently they figured out that my car was one of the cars affected (even though I don't recall telling them about it), and they somehow fixed it from their side. This explains why mine suddenly started working at a specific time and has worked since: it appears they figured out how to fix the problem and fixed mine (remotely). This must have been via software sometime between Apr 10 Monday at 11:32PM (which was the last time I drove it when it didn't work right) and Apr 11 Tuesday at 2:02PM (which was the first time I drove it which it worked right again after many days of not working, and has been working since).

I received no firmware update notices on the car whatsoever, so whatever it was must have been quiet.

My vehicle was also in for 12,500 mile service on Apr 6, before they fixed it remotely on my car, during the period in which it wasn't working at all for me; I didn't tell them about it since I knew the bug was widespread and would eventually be fixed for everyone. The car was returned to me in the same state (the bug still present). It didn't get fixed for ~5 days after service was complete.

My car is parked in wifi at home, and usually has access to LTE when on the road.

I wish someone told you what the name of the bug is, so we can call it the "SDJ92D" bug or something, so we can just call them up and say "hey I have the SDJ923D" bug or whatever, rather than trying to give a random description to Service and hoping they know how to match that to their internal information. But, once any of us with the bug figure out how to describe it to service, it seems like that's all it will take for them to put you in queue to have them root out the bug from your car (remotely).

I messaged Tesla support with a lengthy description of the issue and they referred me to my SC. Subsequently, I have an appointment there on Monday. So if there's a software fix they can push, they chose not to for me.

 

[zonehawk]

So, ever since that Tuesday Apr 11, my AP2 software has been working (today is Saturday Apr 15): speed limit signs, headlamps, lines for the sides of the road, other cars when it can detect them in front, etc.. Whatever problem I had seems to have gone away on its own. I don't know what was causing it. Maybe someone manually logged in to my car from the mothership and cleaned it up, or it reset itself, or a patch came in, or it fixed itself, or the issue causing it has been hidden, or whatever. But, I can say without anything done on my part the problem has not come back since it went away Tuesday.

Ok. It seems that apparently they figured out that my car was one of the cars affected (even though I don't recall telling them about it), and they somehow fixed it from their side. This explains why mine suddenly started working at a specific time and has worked since: it appears they figured out how to fix the problem and fixed mine (remotely). This must have been via software sometime between Apr 10 Monday at 11:32PM (which was the last time I drove it when it didn't work right) and Apr 11 Tuesday at 2:02PM (which was the first time I drove it which it worked right again after many days of not working, and has been working since).

I received no firmware update notices on the car whatsoever, so whatever it was must have been quiet.

My vehicle was also in for 12,500 mile service on Apr 6, before they fixed it remotely on my car, during the period in which it wasn't working at all for me; I didn't tell them about it since I knew the bug was widespread and would eventually be fixed for everyone. The car was returned to me in the same state (the bug still present). It didn't get fixed for ~5 days after service was complete.

My car is parked in wifi at home, and usually has access to LTE when on the road.

I wish someone told you what the name of the bug is, so we can call it the "SDJ92D" bug or something, so we can just call them up and say "hey I have the SDJ923D" bug or whatever, rather than trying to give a random description to Service and hoping they know how to match that to their internal information. But, once any of us with the bug figure out how to describe it to service, it seems like that's all it will take for them to put you in queue to have them root out the bug from your car (remotely).

I wish this was true, but my SC knew nothing of the issue beyond that they hope a software update will fix it.

 

[zonehawk]

Did they attempt camera calibration? Apparently that's why my SC will do Monday.

They did not attempt camera calibration.

 

[Henry169]

The disappearance of AP after 8.1 feels like pure software bug with whatever hardware variations exist in our cars. Hopefully SCs aren't throwing hardware solutions at the problem. My appointment is Monday, I will let y'all know if it's solved.

I have an appointment on Wednesday. The new ECU has been delivered to the SC. I am interested in your experience Monday. I am also a bit worried about the hardware. If my ECU really is broken there must be a reason why it needs to be replaced a second time. If they do not find the root cause I will need a 4th ECU soon :-(

But I am also happy with the idea that on Wednesday I can use AP again

 

[chillaban]

If the software issue only affects a certain batch of ECUs, it's perfectly valid to swap the hardware to an unaffected batch. Or to a yet-to-be-affected one. I deal with situations like this all the time with my line of work -- customer-facing service shouldn't be put in a position where they have to stall customers for an engineering fix, and engineering shouldn't be in a position where they have to rush out fixes or see customers waiting and suffering with under-warranty broken products.

 

[sdorn]

Mine has been working for the past few days again, not sure why. It was not working for quite a while before that. Not sure if it will keep working.

 

[miken79]

If the software issue only affects a certain batch of ECUs, it's perfectly valid to swap the hardware to an unaffected batch. Or to a yet-to-be-affected one. I deal with situations like this all the time with my line of work -- customer-facing service shouldn't be put in a position where they have to stall customers for an engineering fix, and engineering shouldn't be in a position where they have to rush out fixes or see customers waiting and suffering with under-warranty broken products.

Same here, which is why I have hope this is the fix and that they'll have a firmware fix soon as well. They were already in the process of replacing my ECU so I didn't bother to offer to wait again. I've already pushed them to much to fix it, so just accepting the swap. Hoping to have good things to report Monday.

 

[Doubletap67]

If the software issue only affects a certain batch of ECUs, it's perfectly valid to swap the hardware to an unaffected batch. Or to a yet-to-be-affected one. I deal with situations like this all the time with my line of work -- customer-facing service shouldn't be put in a position where they have to stall customers for an engineering fix, and engineering shouldn't be in a position where they have to rush out fixes or see customers waiting and suffering with under-warranty broken products.

My bet is software and out of alignment cameras. SC mentioned camera firmware too. Given all these cars were built in the same few months, I'd be surprised if there are multiple spins of a PCB. I may be wrong, of course.

However and obviously, if they solve it with a brute force hardware fix I don't care if they replace half the car.