Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Model X Crash on US-101 (Mountain View, CA)

This site may earn commission on affiliate links.
yes, it seems the news stories and today's NTSB tweet focus mainly on post-crash and the Tesla-ness of the accident. Presumably the scope of the investigation is much broader and looks into the pre-crash site-specific & traffic aspects too. Anyone know if those dashcam videos and photos of the pre-incident unreset barrier have been forwarded to appropriate investigative bodies?
 
If the car was traveling in the right side HOV lane, there is no probable sequence of events where it should have hit the barrier. That is, unless AP was on, and with adverse light conditions, followed the wrong solid white line, i.e. the one on the left side of the gore point, instead of the line it should have followed on the right side of the gore point. In such a scenario, assuming the driver was not paying attention, AP would have guided the car straight into the barrier at cruising speed.

Yes, the wrong line is actually painted better and less worn than the correct line. I could see a situation happening where driver or autopilot accidentally started to drift to the left following the wrong line.

leftline.png


Also, note, the left most lane is intentionally drifting to the left after the Y split. So if you were trying to stay in the right carpool lane by watching the cars on your left you might be inclined to start drifting left with them.
 
If anyone is able to do some video while on autopilot in that area... trying to change lanes or seeing if the car falters, it would be really helpful. This accident is really awful, I wonder if they will be able to recover the data.

I notice that many similar exits here have the white horizontal lines marking that it is not a lane. So many factors, but the possibility of autopilot being involved has me worried.
There are posts I read where Tesla drivers say the have driven that numerous time without issue. If you are in the off ramp/left lane in this situation I find it hard to believe the car just swerved at the last second that far over to the right and hit the barrier. At this point we don’t know what occurred but from what I have read about that area many cars speed along in the ramp/left lane then cut back onto 101. Others realize at the last second they are actually heading off the highway and cut back onto 101. It is poorly marked as well from the photos of the area I have seen leading up to the MVC.
 
Perhaps they were sued concerning yellow barrels?
We just had a brand new 2 freeway complex interchange system installed. All the yellow barrels are gone. Most gore points have guard rail covers now, and AFAIK, only a single telescoping one.

It's a problem with driving. A road cannot be made idiot proof, because by the time it's finished, they've already developed better idiots to defeat it.
That’s why where I come from they are no longer called “accidents” but “collisions” because an accident is defined as; unfortunate incident that happens unexpectedly and unintentionally. Most collisions are preventable hence the change in wording.
 
Google Maps
line2.png


And with the morning sun in your face (and blasting the auto pilot cameras) it becomes even harder to discern that line.

Also, coincidentally, in that particular picture, a car in that right carpool lane wouldn't see the upcoming gore point wall because the truck in front of them is blocking the view.

Of all the pictures posted so far, this one may sum it up best in terms of what could have gone wrong.
Trying to keep beside the car on your left. Trying to use the more prominent (but wrong) left lane marker, and vehicle in front blocking the view of the hazard ahead.

This is again a scenario where candlestick delineators (plastic poles) in the gore area (space between lanes) would have avoided such a scenario.

Note, the driver would have a better chance of seeing the gore point ahead being more left in the car. The autopilot cameras are centered, so would have the view blocked by the truck longer.
 
Last edited:
Quote from a friend who drives his Teslas through that area:
I’ve used autopilot on that stretch of road and it’s a little scary. The barriers kind of juts out into the roadway and leave a small margin of error. Therefore, I always have my hands on the steering wheel in addition to the autopilot. I hope it isn’t an autopilot issue, as I’d hate to see another setback for that technology.
 
yes, it seems the news stories and today's NTSB tweet focus mainly on post-crash and the Tesla-ness of the accident. Presumably the scope of the investigation is much broader and looks into the pre-crash site-specific & traffic aspects too.

Well their release was in the past tense, which would imply they are done with the investigation and that was the only two things they looked in to. But maybe they left a word or two out...
 
  • Disagree
Reactions: ddkilzer
Yes, the wrong line is actually painted better and less worn than the correct line. I could see a situation happening where driver or autopilot accidentally started to drift to the left following the wrong line.

View attachment 289702

Also, note, the left most lane is intentionally drifting to the left after the Y split. So if you were trying to stay in the right carpool lane by watching the cars on your left you might be inclined to start drifting left with them.
Don’t those two lines you have noted come to a point together further back? If so, drivers paying attention should know this is not a lane. Painting chevrons or diagonal lines through those two lines would have been very helpful.
 
You should be worried. The conversation here has focused mostly on Caltrans responsibility in this fatality, but that's not something you can do anything about in the short term. The real issue for Tesla owners in this crash is: did the car have any responsibility? You can't do anything to change the behaviour of other drivers, or Caltrans right now, but how you operate your car in regards to when you trust Autopilot and when you don't, is something you can change starting right now.

If the car was traveling in the right side HOV lane, there is no probable sequence of events where it should have hit the barrier. That is, unless AP was on, and with adverse light conditions, followed the wrong solid white line, i.e. the one on the left side of the gore point, instead of the line it should have followed on the right side of the gore point. In such a scenario, assuming the driver was not paying attention, AP would have guided the car straight into the barrier at cruising speed.
My reason for worry is not the AP system, Iooking at the images and having considerable experience driving AP2, I doubt the likelihood of Autopilot causing this. No, my worry is that if the data recorder is not available, and in the direct aftermath of Uber-gate, there is a question out there as to whether the vehicle was not on autopilot.... that is very bad potentially. Which gets me back to why the uber situation is such a bad moment in the tech development.

Thoughts to consider, and why I am hopeful about the situation. First of all, it’s awful and tragic that goes without saying, though it still needs to be said over and over again. The road markings and the barrier are the most significant issue barring a complete failure of autopilot (again, something I doubt), and Caltrans is likely more to blame than anything regarding this fatality because any car in this situation would have likely resulted in a fatality. Had the barrier been up, I don’t see a scenario where a buffered concrete barrier causes this sort of damage.

There is a scenario I can imagine where perhaps autopilot was involved. It is possible, that AP was engaged just prior to this event, or an auto lane change was activated when it should not have been available. This would suprise me in an AP2 car where conditions have to be ideal to initiate a lane change.... those of us that drive AP2 a lot know there are situations where it shows the lane change available next to a shoulder. I would be surprised by this on the latest firmware because the lane change is cautious and slow and doesn’t have the diving behavior of past versions.

To me, the most likely scenario at this point appears to be a driver who disengaged autopilot prior to changing lanes and who tried to change roads at the last minute. Regardless I have no doubt that Tesla will use the information collected to render these types of objects and barriers on the IC and within the system. Right now, I keep my fingers crossed that this bad situation is not the worst case scenario where AP drove straight into the barrier somehow. We should also consider the possibility that TACC was engaged and manual steering was the issue. Either way, a last second change seems most likely yes? What do you guys think? I really wish I could test out this road to try and help them figure this out now that this has happened, it would be useful to see if there are any autopilot scenarios that one could replicate at this area of road. Maybe jimmy_d is in the area....
 
And they should just use the huge yellow bucket with sand or water in it, instead of the way more expensive "smart cushion" thing. Yellow bucket with water or sand is just as effective in slowing down the car.

If it is reliably full. However, jokers with a battery-powered drill can mess with it or neglect can leave it drained and useless. And in some cases, metal barriers can be cheaper than the whole big plastic drum and water trucks to fill them.
 
Google Maps
View attachment 289706

And with the morning sun in your face (and blasting the auto pilot cameras) it becomes even harder to discern that line.

Also, coincidentally, in that particular picture, a car in that right carpool lane wouldn't see the upcoming gore point wall because the truck in front of them is blocking the view.

Of all the pictures posted so far, this one may sum it up best in terms of what could have gone wrong.
Trying to keep beside the car on your left. Trying to use the more prominent (but wrong) left lane marker, and vehicle in front blocking the view of the hazard ahead.

This is again a scenario where candlestick delineators (plastic poles) in the gore area (space between lanes) would have avoided such a scenario.

Note, the driver would have a better chance of seeing the gore point ahead being more left in the car. The autopilot cameras are centered, so would have the view blocked by the truck longer.
I see what you are saying on this subject and understand the logic, but my sense of driving AP2 is that autopilot would not have made the error you describe. TACC I could see though. Or even the driver thinking the car is on AP when really only TACC is engaged... it’s one of the reasons why I kind of wish TEsla would use a diffferent ambient color on the IC to represent if the car is on autopilot or TACC, there are times when I am so used to driving on autopilot that I forget that I am only on TACC and realize it when the car is not steering.... if one were to look down at the same time, an accident could occur. I don’t know if a different color, say yellow for TACC and blue for AP, would make a difference in this scenario.... perhaps different tones to indicate TACC from AP, or even different motions of the stalk... rather than pressing something twice... it’s a pull and hold for AP until it engages or a push and hold forward for TACC until it engages... either way a different Audible note would be available..... certainly avoid the phrase bing bing either way.
 
My reason for worry is not the AP system, Iooking at the images and having considerable experience driving AP2, I doubt the likelihood of Autopilot causing this. No, my worry is that if the data recorder is not available, and in the direct aftermath of Uber-gate, there is a question out there as to whether the vehicle was not on autopilot.... that is very bad potentially. Which gets me back to why the uber situation is such a bad moment in the tech development.

Thoughts to consider, and why I am hopeful about the situation. First of all, it’s awful and tragic that goes without saying, though it still needs to be said over and over again. The road markings and the barrier are the most significant issue barring a complete failure of autopilot (again, something I doubt), and Caltrans is likely more to blame than anything regarding this fatality because any car in this situation would have likely resulted in a fatality. Had the barrier been up, I don’t see a scenario where a buffered concrete barrier causes this sort of damage.

There is a scenario I can imagine where perhaps autopilot was involved. It is possible, that AP was engaged just prior to this event, or an auto lane change was activated when it should not have been available. This would suprise me in an AP2 car where conditions have to be ideal to initiate a lane change.... those of us that drive AP2 a lot know there are situations where it shows the lane change available next to a shoulder. I would be surprised by this on the latest firmware because the lane change is cautious and slow and doesn’t have the diving behavior of past versions.

To me, the most likely scenario at this point appears to be a driver who disengaged autopilot prior to changing lanes and who tried to change roads at the last minute. Regardless I have no doubt that Tesla will use the information collected to render these types of objects and barriers on the IC and within the system. Right now, I keep my fingers crossed that this bad situation is not the worst case scenario where AP drove straight into the barrier somehow. We should also consider the possibility that TACC was engaged and manual steering was the issue. Either way, a last second change seems most likely yes? What do you guys think? I really wish I could test out this road to try and help them figure this out now that this has happened, it would be useful to see if there are any autopilot scenarios that one could replicate at this area of road. Maybe jimmy_d is in the area....
I agree with you. It would be interesting to see what firmware this model X was operating under.
 
...I don't think the cut loop removes 12V from the vehicle.

The rationale for disabling the 12V is :

"Its 12 volt battery operates the SRS, airbags, windows, door locks, touchscreen, and interior and exterior lights. The DC-DC converter in the high voltage system charges the 12 volt battery, and the 12 volt battery supplies power to the high voltage contactors, allowing high voltage current to flow into and out of the high voltage battery."

If the 12V battery itself is gone in this collision, there's no low voltage current to power those above components even if I reconnect the First Responder Loop or if I short the 12V wires.

Now that the 12V is gone, there's power to power the DC-DC converter and there would be no 12V coming from that converter either.

Also, for the high voltage, as Electrek pictured for a Model 3: There are multiple redundancies to make sure it got cut off in a collision:

1) Negative Power Switch (I think that's a contactor)
2) Pyro Disconnect
3) Positive Power Switch (I think that's another contactor)


tesla-model-3-battery-pack-modules.jpg
 
If it is reliably full. However, jokers with a battery-powered drill can mess with it or neglect can leave it drained and useless. And in some cases, metal barriers can be cheaper than the whole big plastic drum and water trucks to fill them.
I have seen videos of vehicles hitting those sand and water filled barrels. They make such a huge explosion of water or sand that other drivers are startled and have their view obstructed by it. I believe that is one of the reasons they went away from them.
 
>> A few folks have mentioned that he worked for Apple. I haven't seen a source for that though. Did I miss that? <<
Nothing official. I heard it from someone at work who heard it from someone else.

I have verified (through a mutual friend and former coworker as well as another TMC member at Apple) that Walter Huang did work for Apple and previously for Electronic Arts before that. Note that this doesn't really tell us anything about whether he was heading for Cupertino or Sunnyvale offices (or some other appointment). His family (and investigators) might know, but I don't wish to speculate.

However, I would like to mention that a GoFundMe campaign has been set up for the benefit of Walter's wife and two children, the funds from which will reportedly be managed by his brother: Click here to support RIP Walter Huang organized by Shawn Price
 
There's a direct to module connector for EDR that is somewhere under the passenger seat and under console. From Crash Data Group who sells the Tesla connecting cables:
  • Direct-to-module retrieval: In the Model X and the Model 3, the vehicle’s RCM is located on the floor of the passenger compartment, under the center console and between the seats. In the Model S, the vehicle’s RCM is located on the floor of the passenger compartment, under the center screen.

My husband is usually driving our MS when we travel most of the time. Have used TACC a few times on the highway. If TACC was on, and assume you mean the lane following or car spacing feature was active, wouldn't the car have slowed down even approaching the barrier as it would think it was a stopped car in front of you?
 
  • Disagree
Reactions: RedOctober
Where is the data recorder stored and does @verygreen or @lunitiks or anyone else who knows where this stuff is, think it will be recoverable?
The Event Data Recorder (EDR) looks like it is located underneath the monitor or accessed from there according to a March 6 Electrek picture. Looks like you remove the plastic holder beneath the monitor to get to the access point. Not sure if the EDR is in there or just the access port. There were so much debris, I hope it was saved or collected when they cleaned up the scene.
 

Attachments

  • 2542C45C-796C-4710-9910-5506A9418070.jpeg
    2542C45C-796C-4710-9910-5506A9418070.jpeg
    148 KB · Views: 48
Last edited:
  • Disagree
Reactions: SMAlset
yes, it seems the news stories and today's NTSB tweet focus mainly on post-crash and the Tesla-ness of the accident.
Presumably the scope of the investigation is much broader and looks into the pre-crash site-specific & traffic aspects too.
Anyone know if those dashcam videos and photos of the pre-incident unreset barrier
have been forwarded to appropriate investigative bodies?

I just read the 20+ pages of this thread, but I am still puzzle on how this accident occurred?
- Does the Tesla driver tried to change direction (left or right) at the last moment?
- Does another car was involved?

I have quite often been cut off by drivers changing lane at high speed on the freeway.

- In the situation below, I got cut off by this reckless Jaguar driver, on the Hwy 80 in Emeryville, CA,
while I was driving already at the speed limit.
- I never had a car passing so close to my front bumper.

- I could had also created an accident to avoid this crazy driver, as a stepped on my brake.
- If there was a car behind me, I would have been back ended.

2017 02 25 - BK to SF - Passing Cut Off 02  - Circled .jpg

2017 02 25 - BK to SF - Passing Cut Off 08  - Circled  .jpg

2017 02 25 - BK to SF - Passing Cut Off 09  - Circled .jpg
 
Last edited: