SF Bay Accident Surveillance Footage

Exclusive: Surveillance Footage of Tesla Crash on SF’s Bay Bridge Hours After Elon Musk Announces “Self-Driving” Feature — The Intercept

Musk has said Tesla’s problematic autopilot features are “really the difference between Tesla being worth a lot of money or worth basically zero.”

apple.news

 

[2101Guy]

What's strange is that it's the Bay Area, and so thousands of Teslas have gone through this same path, and likely a bunch of them on AP. What's different here?

some people have reported different FSD/AP experiences even with same model of tesla/same year/same version of software

 

[Matias]

It looks like he might have not responded to the Auto Pilot nags and the car was simply pulling itself over and stopping as its supposed to ? That seems like the simplest explanation.

Regardless, there is something wrong with the driver. He could have just corrected and gotten on the gas to avoid the entire situation.

I have not read this thread or other articles (just watched the video from the rear angle, so this is admittedly just a hot take.

Car would have put hazard lights on.

 

[stopcrazypp]

It looks like he might have not responded to the Auto Pilot nags and the car was simply pulling itself over and stopping as its supposed to ? That seems like the simplest explanation.

Regardless, there is something wrong with the driver. He could have just corrected and gotten on the gas to avoid the entire situation.

I have not read this thread or other articles (just watched the video from the rear angle, so this is admittedly just a hot take.

Mentioned in other thread, but Teslas do not pull to the side when the nag counter is over. All it does is come to a gradual stop in the lane it currently occupies. The reason why is obvious (to avoid exact situations like this where driver in adjacent lane is not prepared to yield).

Edit: I should mention also that there is no emergency lane in the tunnel on that side anyways so it makes zero sense to pull over in a travel lane anyways, even if Teslas had that function, which they don't.

 

[stopcrazypp]

Great points.

SURELY Tesla has logged a TON of data from this accident that would answer a LOT of questions...like they have before when they have posted log data to refute claims in prior accidents where drivers claimed FSD was active.

The only action that Tesla has seemed to have taken in this incident thus far is Elon's censorship on the matter

They have not done so actively for non-fatal US accidents, from what I can find. The last one I can find was the fatal accident in Texas, which Elon commented in a tweet (but completely ignored media requests for comment). No one had serious injuries in this accident and initial reporting as a lot more vague (not as serious accusations of FSD being on like in the Texas accident where police rep claimed no one was in driver's seat). I posted up thread a lot of initial reporting didn't even mention Tesla, much less FSD, likely because a similar huge pileup happened at a nearby bridge just hours before, so media likely chalked it up to just Holiday traffic.

Log posting for non-fatal accidents (without owner request) are for places like China where they still have a PR team. And they only did that after the owner tried to get the media to report negatively on Tesla and actively protesting Tesla. Please post an example where they have done so in the US. I have seen this claimed a bunch of times and every person I responded to failed to provide examples.

 

[stopcrazypp]

Another data point, I see in the comments in one video that NOA blinks six times before changing lanes, and the car blinked only 4 times, so that suggested it was not NOA changing lanes. Can anyone with NOA confirm (I don't have it, I only have regular AP)?

 

[AlanSubie4Life]

But why changing lane manually while still pressing the brake pedal at the same time?

Because the car did it and it was already going that direction.

the comments in one video that NOA blinks six times before changing lanes, and the car blinked only 4 times, so that suggested it was not NOA changing lanes. Can anyone with NOA confirm (I don't have it, I only have regular AP)?

I don’t think this is true. If the user requested a lane change it can happen nearly immediately as long as it is clear. May be some number of blinks for legality (1 or 2) but certainly not 6.

 

[stopcrazypp]

Because the car did it and it was already going that direction.



I don’t think this is true. If the user requested a lane change it can happen nearly immediately as long as it is clear. May be some number of blinks for legality (1 or 2) but certainly not 6.

Presumably, the comment meant a system initiated lane change (in which case it would rule out at least that), as opposed to a user initiated one, but fair point that it doesn't eliminate possibility of user initiated lane change.

 

[Dewg]

Because the car did it and it was already going that direction.



I don’t think this is true. If the user requested a lane change it can happen nearly immediately as long as it is clear. May be some number of blinks for legality (1 or 2) but certainly not 6.

Agreed, when I initiate a lane change it blinks only once or twice before moving. When the car initiates a chance on the freeway, it dings and indicates on screen that it wants to change lanes, asking for a wheel torque to make sure I'm paying attention. Once I satisfy that, it starts the blinker and makes the move. I've not experienced it engaging the blinker for several blinks before moving.

 

[stopcrazypp]

Agreed, when I initiate a lane change it blinks only once or twice before moving. When the car initiates a chance on the freeway, it dings and indicates on screen that it wants to change lanes, asking for a wheel torque to make sure I'm paying attention. Once I satisfy that, it starts the blinker and makes the move. I've not experienced it engaging the blinker for several blinks before moving.

To clarify, this may be talking about the total amount of blinks, not just the blinks before the move. When I have the time I may review the footage to see how many happened (before or after), and maybe look up some NOA footage to see if it is possible to count.

 

[aronth5]

Because the car did it and it was already going that direction.



I don’t think this is true. If the user requested a lane change it can happen nearly immediately as long as it is clear. May be some number of blinks for legality (1 or 2) but certainly not 6.

I agree in general but the number of blinks depends how clear the requested lane change is. Sometimes conditions for the lane move change unexpectedly and you'll get more blinks until NoA/FSD determines the lane change is safe.

 

[AlanSubie4Life]

I agree in general but the number of blinks depends how clear the requested lane change is. Sometimes conditions for the lane move change unexpectedly and you'll get more blinks until NoA/FSD determines the lane change is safe.

Of course. In this case it was wide open, so not an issue. The black SUV was closing pretty fast but only in a range of concern at the time the car started to move over (was a huge two seconds behind - at least - when signal was present).

Presumably, the comment meant a system initiated lane change (in which case it would rule out at least that), as opposed to a user initiated one, but fair point that it doesn't eliminate possibility of user initiated lane change.

Right, I have no idea how it works if the system initiates it but in this case we have no way to know whether it did.

To me it looks like a situation where a user might ask for a lane change - traffic in front is going annoyingly slow, need to get around them. And the Mini looks like a speed demon.

Looks like at least two blinks before the car started to move.

 

[BurgerDave]

Anybody following another vehicle is responsible for stopping without hitting the vehicle in front, period. Having said that, In my experience, Tesla can't even get basic autopilot to run smoothly on the highway without phantom braking a few times a day on an Oregon Coast Highway trip: at the brow of several hills, as it passes a cyclist on the shoulder, when it sees pedestrians well off the highway, when a trades van turns left in front of you when its so far ahead of you that you would not expect braking. In a two lane highway environment, basic autopilot is basically unuseable. No idea how FSD can be trialed in urban driving. My experience with autopilot is that the braking issues are highly disorienting - you don't necessarily understand why its suddenly braking so hard, if you've missed something, or if you should press on regardless...

 

[AlanSubie4Life]

In a two lane highway environment, basic autopilot is basically unuseable.

Yes, AP is not for use on 101. That is a terrible application for it (it’s not even supported there) and manual driving, possibly with judicious TACC use, is what should be used.

Dangerous highway. Drive manually and stay alert.

 

[KLam]

I hope people learn how to safe drive by watching these footage. Especially those involved in the accident.

"Do not follow so close, especially on the left most lane. Extra caution when there are left exits" Your life depends on the people's driving skill before you as you cannot see beyond the car just in front of you. Stay away when you see people tailgate in front of you.

 

[brkaus]

@KLam i have my kids watch some of the videos here. Good way to see how drivers act and what to expect.

 

[BurgerDave]

Yes, AP is not for use on 101. That is a terrible application for it (it’s not even supported there) and manual driving, possibly with judicious TACC use, is what should be used.

Dangerous highway. Drive manually and stay a

Yes, AP is not for use on 101. That is a terrible application for it (it’s not even supported there) and manual driving, possibly with judicious TACC use, is what should be used.

Dangerous highway. Drive manually and stay alert.

I shouldn't have said autopilot, I should have said: Traffic-Aware Cruise Control. We don't use the steering portion of the autopilot system as that is waaay too dangerous!

​

 

[AlanSubie4Life]

Yeah even TACC use would be pretty questionable there. Ok judiciously, on the easier sections. But will definitely open you up to phantom braking in a lot of places.

Doesn’t seem like the most likely explanation of the Yuerba Buena issue though. Who knows.

 

[Scott Chapman]

I do have some phantom braking with my '22 MSLR, but it is MUCH better than my previous '18 Audi S5 which pretty much did the same thing...

 

[KLam]

Ok, after watching the clips a lot of times and based on my experience with NOAP. This is my guess.

The Tesla driver put on the left signal. NOAP initially showed a slot on the blue left destination lane and tried to change lane. But, the last moment it detected the fast approaching Black SUV (the slot turned to red) and started to abort to maintain the current lane. However, the driver tried to hold onto the left turning steering wheel (yoke). This disengaged the NOAP and sounded the Red Wheel Collision Warning and put on the brake. Driver was startled by the Loud alarm and do not know what to do and might have put further braking which stopped the car. (from the clip, I sensed the slight hesitation on the left lane change and struggle)

I have to say I encounter this last minute abort before. So now I don't ask the system to left lane change when there are cars on my left close behind. Or, I will 2 taps on the brake (2 because I want to make sure it's not still on cruise control) to disengage and do the lane change myself.

 

[clydeiii]

New theory:

Background: on my morning daily commute I have a bunch of different options. I can take the highway all the way to my work or get off on a bunch of different exists and take city streets. Because my work has a few paths of entry, some of which are closed at certain times during the day, I can't wholly rely on the car's navigation and Google's traffic reports to get me to the right entrance at the right time (since it has no idea about the various gate hours).

So very often the car will want to get off the highway early and drive on city streets into my work but I know that's a bad idea, because the gate it wants to drive towards will be closed, so I force the car into the left (passing) lane and turn off NoA while keeping AP engaged (via on the on-screen NoA button), so that the car definitely won't take the exit off the highway. And when I do this, 100% of the time the car WILL SLOW DOWN, thinking that it's missing its exit, and wanting to somehow correct the error. Since I know this is standard NoA/AP behavior (and definitely NOT FSD behavior, 2101Guy!--totally different thing), I am ready to lay on the accelerator or sometimes I just disengage AP altogether, depending on my mood.

So, with that background, here's the theory:

We all know that sometimes when you enter into tunnels, the GPS unit can get confused and put you onto a totally different part of the map than you're actually on--standard GPS unit errors. Maybe the driver had NoA on and upon entering the tunnel, the GPS unit got confused about where it was and felt that, in order to keep on the right path, it needed to take an immedate exit on the right. But then it believed it was going to miss that exit, so it felt the need to slow down, similar to my car on the highway when it's missing an exit. And thus why we see it slow AND try to change langes at the same time.