I was skeptical...

[Saghost]

TACC does turn off when you hit the breaks. So, that condition already exists. I have observed the following conditions:

• TACC will accelerate if the lead car slows below your speed setting and falls from view via radar or camera
  • Most common and personally experienced when making a 90 degree turn while following without disengaging
• TACC will disengage when break is applied
• TACC can only be engaged:
  • Above 18 MPH
  • At stop with lead vehicle detected and break hold engaged. I also noticed that it wouldn't engage until my foot was off the break and in break hold. I was unable to replicate conditions outside of these conditions including some of the conditions stated by posters on this thread.

I do get how a user could either inadvertently engage TACC or forget it is engaged creating potential unsafe conditions within the design specs. But, the car is operating as designed. The user created the unsafe condition inadvertently. My previous vehicle, Buick, had a similar feature and behaved in the same unsafe manner with the same conditions. At what point does the user take responsibility for creating the unsafe condition inadvertently or otherwise. IMO there is not an issue with the firmware, bug, that is causing the conditions. I would recommend that you take the time to understand the functionality and conditions where you need to manually intervene to avoid sudden acceleration. I would recommend practicing a bit with the TACC behaviors in safer environments before using frequently.

To state the obvious if you didn't look at my signature, I am on AP2.0 firmware version 17.17.4.

Brake hold isn't relevant to whether you can engage TACC/AP at low speeds - you just have to be off the brake pedal with a car in front.

If you have creep turned off, you can sit at zero mph and engage them without brake hold. With creep on, you'll get accelerated to five mph or so just by releasing the brake, but still be able to engage TACC/AP as soon as you're completely off the brake if there's a car in front somewhere.

They can't allow engagement while you're on the brake pedal because being on the brake pedal disengages both...

 

[Saghost]

A simple solution to ALOT of this would be to simply code that ALL AUTOMATIC ACCELERATION IS DONE AT A MODERATE OR EVEN GRANNY PACE.

Heck... There should be a "granny" mode in addition to Ludicrous. No offense to grandmothers.

I don't think I agree. There's already a limit in place which seems to be the main reason TACC doesn't manage your following distance well and people cut in on you during freeway stop and go situations at low following distance settings.

The car would be better for freeway driving if this limit was raised or eliminated, lowering it will make things painful.

Maybe a smarter limit would work - limit depends on TACC distance setting and is different when following a car than not, or different on freeways from city streets?

 

[AnxietyRanger]

@Saghost But the logic in disabling TACC whenever AP is disabled comes from reveresing the activation logic.

Now, you can enable both TACC and AP at the same time - which causes an audible bong - but when you take over steering, you hear that audible bong again, but only AP disengages, leaving TACC on.

Basically from no mode you go to mode A, but disabling mode A doesn't take you back to no mode at all, but it takes you to mode B instead. Clearly grounds for mode confusion.

 

[habanero69]

To the OP:
......................
I hope, in future, drivers of all cars can download the logs right away, before the manufacturers get any chance to tamper with the logs. This will be a very common point of litigation in the future, when semi autonomous cars become main stream.

I can't see this happening. You already have no shot at getting logs from ICE cars (unless your attorney subpoenas them, and you still don't know if they've been tampered with). Plus, they may in the future encrypt so that if you were able to get on your own, they'd be un-readable. I'm not saying this is my want or position, but just how I see us being able to get the logs our self play out... YMMV.

 

[AnxietyRanger]

I can't see this happening. You already have no shot at getting logs from ICE cars (unless your attorney subpoenas them, and you still don't know if they've been tampered with). Plus, they may in the future encrypt so that if you were able to get on your own, they'd be un-readable. I'm not saying this is my want or position, but just how I see us being able to get the logs our self play out... YMMV.

This could be legislated, though.

 

[habanero69]

I suppose "anythings possible" when you have software control acceleration and braking. Maybe someone left out a semi colon at end of the a line of code or something.

I wonder if OP can try to park at same spot and see if same thing can be reproduced.

How many have reported something like this?

I know you probably are poking fun with the coding comment, but it's not possible. Either the code routine isn't written (properly) to process the inputs it received, OR, the inputs (from a sensor for example) were totally incorrect for the actual situation. Either could be called a bug, but for the person that experiences it, it is just a "fail"...

 

[habanero69]

This could be legislated, though.

You are talking about our house/senate? That will be the day they put us, the citizen, ahead of companies...

 

[Saghost]

@Saghost But the logic in disabling TACC whenever AP is disabled comes from reveresing the activation logic.

Now, you can enable both TACC and AP at the same time - which causes an audible bong - but when you take over steering, you hear that audible bong again, but only AP disengages, leaving TACC on.

Basically from no mode you go to mode A, but disabling mode A doesn't take you back to no mode at all, but it takes you to mode B instead. Clearly grounds for mode confusion.

???

That's not a point I was attempting to debate today.

You can activate TACC without Autosteer, too - and then come back later and activate Autosteer without dropping out of TACC.

TACC without Autosteer activates with only the blue highlight to indicate it is active just like it stays active with only the highlight.

The system behavior is logical and internally consistent - and the visual cues TACC includes are more prominent than most cruise controls I've driven with over the years.

I'll agree there could be potential for confusion; I'm not sure there's an obvious way to prevent the confusion short of making Autopilot all or nothing - not allowing TACC without Autosteer, and then using the audible prompt for both.

As long as Autosteer continues to be reluctant to change lanes at times and requires correction at others, I wouldn't want my use of the steering wheel to cause TACC to drop on freeways.

 

[AnxietyRanger]

That's not a point I was attempting to debate today.

Fair enough. My takeaway from your posts were that you were against disabling TACC when taking over steering from AP. If I misunderstood your position, apologies.

I understood that to be one of @bonnie's suggestions (disable TACC too when taking over from AP) and I agree with that suggestion, because it would simply lessen move confusion. Hearing that bong again is a sign to take over in every way...

I know you can enable TACC first, then AP, in which case disabling AP only when taking over steering makes a bit more sense. I actually did this while driving today. But I think the system should err on the side of caution.

If AP disengages, it could alternatively also give the driver some time while TACC remains active before disengaging that as well.

The system behavior is logical and internally consistent - and the visual cues TACC includes are more prominent than most cruise controls I've driven with over the years.

I find the visual cues quite mild. AP2 does not highlight the leading car at all compared to TACC (no blue lead car - at least not here in Europe), so dark vs. white lead car is a very small cue. The blue circles on top are not that visual either and can be seen there even when not active, though in a more subdued color...

In my Audis, which I've driven several generations of adaptive cruises, the bright green symbols have been very distinct and most importantly disappear when not active. And of course other manufacturers looking at self-driving are planning very visible green lights in the dash/steering wheel to show when autonomous is active...

Tesla uses fairly mild visuals and even audio, its bongs and beeps are often very soft... Maybe they really should make this even more fool-proof...

 

[habanero69]

In 1984, when I was learning to drive, upon gently accelerating from stop after a green light, the spring of the accelerator pedal of our 1976 VW Beetle broke off and the accelerator stuck to the floor. Pushed brake and clutch and turned motor off in time luckily to avoid car in front of me. So these things did happen indeed. I suppose there must, today, still be a spring (or perhaps more than one spring) in the accelerator pedal, but I'm 100% confident there are also sensors in there which wouldn't let that happen in a 2017 car.

In a Tesla, I would imagine it is drive by wire for the accelerator. If not, I would be

 

[habanero69]

???

................................

As long as Autosteer continues to be reluctant to change lanes at times and requires correction at others, I wouldn't want my use of the steering wheel to cause TACC to drop on freeways.

^^^^^^^^^^^^^^^^^^^ 100% AGREE.

 

[McRat]

I don't think I agree. There's already a limit in place which seems to be the main reason TACC doesn't manage your following distance well and people cut in on you during freeway stop and go situations at low following distance settings.

The car would be better for freeway driving if this limit was raised or eliminated, lowering it will make things painful.

Maybe a smarter limit would work - limit depends on TACC distance setting and is different when following a car than not, or different on freeways from city streets?

If you are trained, 18 years old, and in the top 10% of your group, drive 2 footed, you can react to brake lights in 2 car lengths at 70 mph if focused and ready for it.

The correct following distance to avoid hitting things with your car will always allow "people to cut in on you".

So prepare for people to change into your lane in front of you, or prepare for an airbag in the face. Many people chose the airbag to the face, hence the crashes we see every day on simplest of all roads, the freeways.

 

[Saghost]

Fair enough. My takeaway from your posts were that you were against disabling TACC when taking over steering from AP. If I misunderstood your position, apologies.

I understood that to be one of @bonnie's suggestions (disable TACC too when taking over from AP) and I agree with that suggestion, because it would simply lessen move confusion. Hearing that bong again is a sign to take over in every way...

I know you can enable TACC first, then AP, in which case disabling AP only when taking over steering makes a bit more sense. I actually did this while driving today. But I think the system should err on the side of caution.

If AP disengages, it could alternatively also give the driver some time while TACC remains active before disengaging that as well.



I find the visual cues quite mild. AP2 does not highlight the leading car at all compared to TACC (no blue lead car - at least not here in Europe), so dark vs. white lead car is a very small cue. The blue circles on top are not that visual either and can be seen there even when not active, though in a more subdued color...

In my Audis, which I've driven several generations of adaptive cruises, the bright green symbols have been very distinct and most importantly disappear when not active. And of course other manufacturers looking at self-driving are planning very visible green lights in the dash/steering wheel to show when autonomous is active...

Tesla uses fairly mild visuals and even audio, its bongs and beeps are often very soft... Maybe they really should make this even more fool-proof...

Actually, I think I mostly am against TACC disabling on a wheel pull, at least on the freeway, as I attempted to explain in the follow-up - that just wasn't the subject of anything I said recently, which left me confused.

A 30s timer would cover my typical use case for TACC after disabling Autosteer, as would a freeway geotagged behavior change. I'm not sure I like either one, though, because both are introducing unexpected inconsistent behavior for the user.

 

[Saghost]

If you are trained, 18 years old, and in the top 10% of your group, drive 2 footed, you can react to brake lights in 2 car lengths at 70 mph if focused and ready for it.

The correct following distance to avoid hitting things with your car will always allow "people to cut in on you".

So prepare for people to change into your lane in front of you, or prepare for an airbag in the face. Many people chose the airbag to the face, hence the crashes we see every day on simplest of all roads, the freeways.

Perhaps. But the distance that TACC maintains on deceleration or in steady state travel is much shorter than the opening it creates during acceleration.

You'd think deceleration would be where your little lecture would apply most...

 

[habanero69]

The issue is not whether TACC turns off or not when you touch the brakes. The issue is that when you take control of the wheel, you are notified that AP is off, but TACC does NOT turn off under that condition. So 1) brakes touched, AP + TACC disengage, 2) wheel turned, AP only is disengaged.

I purposely talked up thread about 'use error' vs. 'user error' because the former doesn't cast blame and encourages looking for the root cause. It's a design failure when the only way to prevent a use error is by 'we put it in the manual'. That's fine for kitchen appliances, but when it's a medical device (my field), a plane, or a moving vehicle, understanding what a user believes should happen is a very important part of the design (and indeed, with medical devices at least, it is required by law to understand what your defined user base believes will happen). Usually user decisions are being made in a compressed amount of time & anything the designed UI can do to encourage the user to make the right decision is a sign of good design.

Good design takes into account what the user believes will happen. In this case, you have two very different behaviors when you exercise the two different ways to disengage AP. The behaviors should be the same. Disengaging AP should either disengage TACC each time or should not. But disengaging only when brakes are touched but not when the wheel is turned is a UI problem. It's an introduced inconsistency.

Eloquently AND clearly stated.

 

[rush6410]

I agree with @bonnie. A simple improvement would be that taking control of the steering wheel would disable both AP and TACC.

The second improvement I'd suggest is removal of cruise control activation by moving the lever up or down, which seems to have the potential to be confused with the blinker leading to . Activate cruise by a short pull and AP by a long pull of the lever and leave the up and down movements to speed changes while activated (and no effect while unactivated).

Third, adding a Settings option to disable cruise control stalk would seem like a good idea for those that want to be sure.

Finally, to avoid mode confusion, improvements could and should be made so that user is always better aware of TACC or AP activation. I would add separate green icons for them that appear when active and disappear entirely when they are not active. Or maybe they could even color the entire instrument cluster edge with a different color based on TACC/AP mdoe? Just coloring of circles blue that are always there does not seem so effective IMO....

I appreciate the conversation to improve the functions. The use of a stalk instead of steering wheel controls for cruise and placement close to turn signal can be confusing. I just don't think it reaches the level of unsafe to a point of liability or legal action as compared to the industry. That is what I was challenging.

 

[bak_phy]

I was bothered with my first Model X drives when I accidentally used the CC stalk for turn indication and the vehicle would speed up and instantly activate cruise control in a parking lot! Why Tesla turns on the cruise with an up/down change speed control is beyond my comprehension. IMHO, the cruise should only be activated with a pull on the lever.

I've done this a couple of times. I'm pretty sure that I touched the CC stalk as I turned the wheel. As surprising as the sudden acceleration is, it's nothing at all like flooring it and I always had plenty of time to correct. If the car has gone to full acceleration there is no way that I would not be in a ditch or worse. I have a MS P90DL.
I could be wrong but it seems that most of the sudden acceleration issue have been with P cars. I think that people are underestimating just how fast they can get into problems when hitting the wrong pedal.

 

[animorph]

Perhaps. But the distance that TACC maintains on deceleration or in steady state travel is much shorter than the opening it creates during acceleration.

You'd think deceleration would be where your little lecture would apply most...

I found the acceleration to be adequate, but there is a second or two delay before acceleration starts.

 

[rush6410]

If you have creep turned off, you can sit at zero mph and engage them without brake hold. With creep on, you'll get accelerated to five mph or so just by releasing the brake, but still be able to engage TACC/AP as soon as you're completely off the brake if there's a car in front somewhere.

Fair enough. I do not use or have experience with creep turned on. I'll add that as a mental note if I ever do turn creep on.

 

[bushburner]

Here is the update. The ranger looked at the logs the other day and didn't see anything out the norm, but to make sure he contacted the regional tesla engineer. The engineer concluded the same thing, and said the throttle never exceeded 25%. Guess I was an idiot for thinking the car had accelerated on its own.

After hearing this, here is my explanation. I turned the wheel left to drive up the angled curb and as I turned the wheel right to straighten out the car, the car moved abruptly due to the weight of the car and gravity. At that moment, I thought the car accelerated on its own and I was the victim of sudden unintended acceleration. Turns out I was wrong and I apologize for causing any concerns. I was skeptical before and now I will renew my skepticism.

Can a moderator update the original post to include the update?