Cruise control no longer safe

[Pagemakers]

I know there are plenty of threads on phantom breaking but having just done another motorway trip I believe that cruise control is no longer safe to use.

The purpose of this thread is two fold.

1) To share something I noticed to see if others have seen something similar, and to see if that helps 'whoever' to fix it.

2) How to get Tesla to recognise the issue and fix it?

On the outbound journey, every time I passed large lorry on my left it braked, on one occasion a white van was very close to going into the back of me.

I did notice that the on screen depiction of the lorries was very close, closer than they were in real life.

The motorway had three lanes

On approaching the next lorry (it was in the far left lane) I moved to the far right lane, to overtake it and no problem!

However, I noticed that while I was overtaking the lorry, its depiction on the screen moved from the far left lane to the middle lane and went red as I was overtaking (the lorry itself remained in the far left Lane). Having noticed this it happened quite a lot

Is this the underpinning bug / has anyone else noticed this???

I did a hard reboot (break pedal and both buttons on the steering wheel), but the problem remained on the return journey

I am on 2020.48.10

How can we get Tesla to recognise this and priorities a fix as it genuinly does not now feel safe to use

Thanks

Like the op I have this issue 100% as described.

I can replicate it every time.

1. Autopilot on
2. Lorry (not car) in the slow lane
3. You in the fast lane
4. When you pass the lorry the display shows the lorry momentarily jumping from the slow lane to the middle lane and back again
5. If you are in the process of changing lanes when this happens you car will VIOLENTLY swerve back into the fast lane and/or VIOLENTLY brake hard. If somebody is passing you in the fast lane at the time a collision is very possible.

I have reported this to Tesla and have had a long and continuing written discussion with them. I have invited them for a test drive in my car or a car of their choosing. The issue can be demonstrated with 100% confidence that the car will exhibit this very dangerous behaviour.

Here is a video I made myself a few weeks ago.

What is interesting is the Tesla only shows traffic in the 2 lanes of the motorway not 3. In sure it was never like this before. Maybe that is the problem here.

 

[Gasman23]

What conditions to you notice it better/worse than others?

I'm not aware that things get recalibrated after an update, but almost certainly functionality changes from release to release as Tesla change and/or compare operation from one version to another even though no mention in the release notes.

I’ve just put it down to weather conditions but haven’t been logging it to compare. It’s just I notice than on days it’s not working well I’ll get a load of phantom brakes and twitchy behaviour. On other days it works flawlessly. I’ve got an update to do now so I’ll see if that makes a difference over the first few drives

 

[Bobbins]

I’ve had my M3 for 8 months and after a couple of months of despair with the TACC I gave up and rarely use it now, it’s less stressful that way. Each time I do venture back into the TACC lion’s cage very quickly it lets me know nothing has changed so I stop using it again. I do 25K miles/year and a simple non-TA cruise control would be a 10 fold improvement, as it is I consider it basically too dangerous to use.

 

[Battpower]

Each time I do venture back into the TACC lion’s cage very quickly it lets me know nothing has changed

Same here. What troubles me is that over several updates there is no sign of consistency or solid improvement. Sure, I have made journeys of 1-200 miles where I have been left feeling things were sorted, only to experience harsh phantom braking the very next trip.

This is even more of an annoyance when additional features like FSD are rendered of little use as a consequence.

Old FSD and City FSD are really amazing imo in terms of what they can already do, but at the same time staggeringly deficient in some basic things they appear unable to do.

 

[Jarovan]

I have never had any problems with cruise control, fortunately, however all your comments seem to be pretty scaring for me now that it might happen one day

 

[Battpower]

I have never had any problems with cruise control, fortunately, however all your comments seem to be pretty scaring for me now that it might happen one day

These are mainly UK owners commenting here I think. Having lived and driven in The US for 15 or more years, many roads layouts are different. Also there are differences in markings, lane conventions, humidity, driving styles etc.

I am convinced there are factors that could effect all Tesla models and in all locations. It's good to be aware while being happy that your use seems uneffected.

 

[pdk42]

My Model 3 TACC experiences are pretty similar to most of the posts here. It’s never been good, and I’m always wary of using it. However, this hard braking when passing lorries is not something that the system did when I got the car. Can’t say when it changed, but it’s definitely worse in that regard.

But it’s not the only safety-compromising behaviour. On the A46 near Stratford for example, a wide 2 lane A road, there’s one section that has a road on the left and a filter lane on the right for traffic coming in the other direction to turn into it. AP will violently pull the car right and into the filter lane since it thinks that’s where the lane is.

It does seem to me that the system is very basically flawed. Maybe it can be fixed with better software, but it’s possible that the camera and radar hardware just can’t get reliable enough data.

Finally, I’ll just add that whilst the Tesla system is probably the worst of these driver automation systems out there, the others have their problems too. My wife recently got an ID3 which has TACC, Forward Braking Assistance, and Lane Keeping Assist. The TACC is way more reliable than the Tesla version and I’m yet to see it do anything even vaguely odd on a highway. However, the lane assist on typical UK streets will try to steer you into parked cars in its effort to keep you in the lane, and the front braking assist can get confused with parked cars on roads where there is a slight bend to the right - in other words it thinks you’re going to collide with the cars even though the road takes you away from them. Thankfully the forward braking assist can be turned down, or even turned off permanently, so it’s not a big deal.

Personally, at the level they’re at I think all these fancy automation systems are more trouble than they’re worth.

 

[MarioG]

Like the op I have this issue 100% as described.

I can replicate it every time.

1. Autopilot on
2. Lorry (not car) in the slow lane
3. You in the fast lane
4. When you pass the lorry the display shows the lorry momentarily jumping from the slow lane to the middle lane and back again
5. If you are in the process of changing lanes when this happens you car will VIOLENTLY swerve back into the fast lane and/or VIOLENTLY brake hard. If somebody is passing you in the fast lane at the time a collision is very possible.

I have reported this to Tesla and have had a long and continuing written discussion with them. I have invited them for a test drive in my car or a car of their choosing. The issue can be demonstrated with 100% confidence that the car will exhibit this very dangerous behaviour.

Here is a video I made myself a few weeks ago.

What is interesting is the Tesla only shows traffic in the 2 lanes of the motorway not 3. In sure it was never like this before. Maybe that is the problem here.

I believe what you show here is (one of) the route of the problem. The sensors detect the shadows and they interpret those as vehicles, that's why the phantom breaking. In both of your cases where two phantom lorries appear in the window, in the first one it is the shadow from a huge tree + real lorry on your left and the second phantom lorry appears due to the shadow from the bridge.

 

[Gasman23]

My Model 3 TACC experiences are pretty similar to most of the posts here. It’s never been good, and I’m always wary of using it. However, this hard braking when passing lorries is not something that the system did when I got the car. Can’t say when it changed, but it’s definitely worse in that regard.

Similarly it’s got worse, making my EAP purchase look like a con. As Tesla can’t seem to solve the issue despite being aware of it, they should at least offer us the option of rolling back to an older software version

 

[Yachtsman]

Same here. What troubles me is that over several updates there is no sign of consistency or solid improvement. Sure, I have made journeys of 1-200 miles where I have been left feeling things were sorted, only to experience harsh phantom braking the very next trip.

This is even more of an annoyance when additional features like FSD are rendered of little use as a consequence.

Old FSD and City FSD are really amazing imo in terms of what they can already do, but at the same time staggeringly deficient in some basic things they appear unable to do.

Yes but he's improved the games with the latest update.

 

[Matias]

I'm not sure about the straps, but I'm 100% with you about it being curtain siders that are the most predictable at causing phantom braking, especially when there is no lane changing involved.

Also 100% that there is some underlying fundamental and dangerous behaviour that is not getting resolved.

I reckon I can predict 30-50% of phantom braking on motorways. My current most likely trigger is a full length artic, white with no signage or writing, slightly loose / flapping siding.

The other related issue I find very suspicious links back to green's early (and subsequent) posts showing clouds of radar reflections from a variety of objects that are not actually in front of the radar path.

My opinion is that this is secondary radar reflections, especially when linked to over head signs and passing under bridges.

Each trigger has very specific causes / conditions, and the only one I so far disagree with is shadows. I do not think the shadow is relevant, but the object casting the shadow could be.

Objects jumping around and suddenly changing type / orientation is a concern if the visualisation we see actually has any correlation with what the car bases decisions on. The other day I pulled up along side quite a high wall and the visualisation turned it into an intermittent large lorry!

I've followed so many threads about old 'fake' fsd and new 'real' fsd, 4D rewrites etc, and the one thing I was hoping for was that visualisations would become stable / non jittery. I do not understand how a safe system can be prepared to habitually and consistently be in a state of uncertainty / flux about what it is seeing. By the time you are along side a truck, your system logic should tell you that truck can't step directly sideways into your path.

My understanding was that by introducing stitched video from all sources and tracking objects in that combined feed a lot of random impossibilities would be eliminated.

I suppose there is a conflict between knowing certainty based on analysis and prediction compared with an ability to respond to the unpredictable. At the beginning, a NN can believe anything, including the completely impossible. It seems that even with a lot of training, very thin threads of the impossible still remain.

Maybe it is the side underrun protection in trucks , which is mandatory in EU, but not in the U.S..

 

[Battpower]

Maybe it is the side underrun protection in trucks , which is mandatory in EU, but not in the U.S..

That is a difference, but not sure how that would confuse the car and also it isn't specific to curtain sided vehicles.

With so many mentionning shadows (often combined with an object overhead) I think some visual element can play a part in braking events, but I am certain that there is a category of event that is common with nothing more than a curtain sided truck.

There are other situations where vehicles are visualized with incorrect sideways) motion, possibly due to switching between camera images. Not sure if these cause phantom braking though.

 

[gdallas]

Not sure if this has been brought up in the 5 or so pages thus far, but it's my understanding that a batch of the 2021s were shipped with front left cam detached because the harness was made too small. Certainly loads of people complaining about it. My 2021 was shipped with this defect. Presume this lack of cam can't be helping the performance of autopilot capability. Juts a thought

 

[gangzoom]

^ Phantom braking is a software implementation. I use to quite happy using AP to drive pretty much the whole way from Leicester to Cardiff, but now I don't even use it for empty dual carriage ways, its simply become too unpredictable.

 

[gdallas]

^

"Measurement is the first step that leads to control and eventually to improvement. If you can’t measure something, you can’t understand it. If you can’t understand it, you can’t control it. If you can’t control it, you can’t improve it,” said American businessman, author, and improvement guru, Dr. H. James Harrington"

As a control and instrument engineer myself, all I was suggesting was that the lack of front left cam may possibly be hindering the control aspects of autopilot and hence phantom braking may be more prevalent because of this in the recent 2021 batches.

 

[pghstochaj]

^

"Measurement is the first step that leads to control and eventually to improvement. If you can’t measure something, you can’t understand it. If you can’t understand it, you can’t control it. If you can’t control it, you can’t improve it,” said American businessman, author, and improvement guru, Dr. H. James Harrington"

As a control and instrument engineer myself, all I was suggesting was that the lack of front left cam may possibly be hindering the control aspects of autopilot and hence phantom braking may be more prevalent because of this in the recent 2021 batches.

I don’t think it can turn on with a faulty camera?

 

[Glan gluaisne]

Cruise control doesn't seem related to AP problems, AFAICS. I've never has a faulty camera, yet cruise is consistently poor, with seemingly random braking events, often very severe (severe enough to risk getting rear-ended, and seriously frighten passengers).

 

[gdallas]

I don’t think it can turn on with a faulty camera?

It does mate. Mine does anyway. Front left been goosed since pick up on 17th. But auto pilot or cruise can be enabled whenever. It works for the most part tbh but I keep my foot close to the accelerator incase it slams on the brakes randomly lol.

 

[Battpower]

Cruise control doesn't seem related to AP problems

My thought is that the same mechanism for maintaining distance and stopping rapidly under some circumstances when using AP is the same as for TACC. The main difference with AP being auto steering and lane positioning.

I'm not sure, but thought only some of the cameras are required for AP..... Like 4.

 

[MrBadger]

I'm not sure, but thought only some of the cameras are required for AP..... Like 4.

Last I saw, pre any confirmed 4D sensor fusion changes that may or may not have happened, the rear facing front wing indicator cameras are only required for NoA/auto lane change, part of the FSD package. They are used for a couple of non autopilot things such as, I believe blind spot detection (that may be primarily ultrasonics) and the new reversing view. This may explain why some reports of camera repairs being prioritised for FSD cars.

So unless you have FSD, the misfunctioning front indicator cameras should not affect TACC/autopilot performance.