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

[TEG]

Something else I had been pondering about this is if the crumple beams in the front of the X managed to miss hitting the smart cushion. Based on the observed damage, I think it may be possible that the left crumple beam may have gone right into the center of the smart cushion and the right one didn't hit anything. In that case, with a pre-collapsed smart cushion, and Tesla crumple beams possibly missing the smart cushion, there would basically be no slowing of the vehicle at all before impact forces caused massive deceleration forces on the driver.

-----

I think the government crash tests tend to run vehicles into wider barriers with no holes / channels in them.

 

[smac]

EAP's performance does not mean that the FSD development is not going well. Only that is has not been integrated into the released version. Which is itself an interesting question: ultimately how will Tesla differentiate FSD from EAP? I don't see EAP having less collision avoidance than FSD. EAP will definitely require a driver in the car, but what else? Steering wheel nags? Driver attention monitor camera?

For level 3 it needs to shift responsibility from driver to car whilst operating within certain constraints. This is a subtle point, but if we take the Audi A8's advertised specs, they are indeed allowing you to read texts / watch movies etc (maybe eating your breakfast is too far, as suggested by an earlier poster) while the car is driving itself.

Now what the Audi will not do is drive you on any roads that don't have a central divider, or operate above 37mph, or without a lead car in front. The system rigorously enforces correct use via capacitive sensors needing BOTH hands on the wheel and ignoring the hand over request immediately puts the car into the equivalent of limp mode once its uncertainty of safety becomes too great, eventually coming to a complete stop.

Now one could argue sensing hands on wheel is somewhat meaningless in determining driver attentiveness (after all you can fool the system by resting one on the wheel in the Tesla implementation, while still looking at your phone for extended periods). However what it does do is prevent the most likely cause of driver inattentiveness, i.e. messing with a mobile phone while driving.

Where I would like to see Tesla get to is Level 3 with the current EAP fleet, and Level 4 (Geofenced) or 5 with FSD.

Personally I think 3, 4 & 5 are where the bulk of accident reduction will come from. (And 4 and 5 have real benefits in traffic management on top of 3). I think the accident reduction from reduced fatigue does come into play, but realistically only on long journeys. For a typical rush hour commute, probably not so much.

 

[PhantomX]

Something else I had been pondering about this is if the crumple beams in the front of the X managed to miss hitting the smart cushion. Based on the observed damage, I think it may be possible that the left crumple beam may have gone right into the center of the smart cushion and the right one didn't hit anything. In that case, with a pre-collapsed smart cushion, and Tesla crumple beams possibly missing the smart cushion, there would basically be no slowing of the vehicle at all before impact forces caused massive deceleration forces on the driver.

View attachment 291716
View attachment 291717

But how does the impact cleanly shear off both beams, the cross braces above in front of dash, and the aft joints of front quarter panels? Both of the beams you pointed out are designed to carry loads in forward and aft direction. If those beams were damaged enough to break away from the car, they would also have transferred enough loads into the main cabin to cause more damages than shown in car. Unless the loads that damaged them were perpendicular to the beams. Same with the cross members and the quarter panels. They are all meant to transfer loads continuously into the rest of unibody structure to distribute impact loads. I keep on asking this because the damage makes no sense as result of head on collision to the barrier unless Tesla really did design front end to break away at specifically energy level.

 

[mongo]

But how does the impact cleanly shear off both beams, the cross braces above in front of dash, and the aft joints of front quarter panels? Both of the beams you pointed out are designed to carry loads in forward and aft direction. If those beams were damaged enough to break away from the car, they would also have transferred enough loads into the main cabin to cause more damages than shown in car. Unless the loads that damaged them were perpendicular to the beams. Same with the cross members and the quarter panels. They are all meant to transfer loads continuously into the rest of unibody structure to distribute impact loads. I keep on asking this because the damage makes no sense as result of head on collision to the barrier unless Tesla really did design front end to break away at specifically energy level.

With the left tire into the left barrier vertical, the left car beam into the void, and the battery pack into the right vertical, the right beam would miss the barrier entirely. The right barrier vertical would effectively push the front cross bumper bar toward the cabin followed by the front motor carrier, both of which would impart a huge side load onto the main side car beams bending them toward the center line and potentially partially fracturing them at the firewall. Follow that with the impact from the car that hit the X in the #2 lane, and the weakened structure could be removed entirely.

Here is a shot of the undercarriage during impact with a flat obstacle.

 

[22522]

I think one of the reasons why the gore area is so long there is that it was originally designed to have a much longer crash barrier in front of the cement wall.
This was the original barrier there:
View attachment 291713

Thank you. That is longer, but still a mysterious object when over a football field away...

CALTRANS should at least add a bicycle flag.

 

[MXWing]

Everyone keeps suggesting that he was distracted by his cell phone. We don't know that. What if in those crucial 5 seconds he was checking his mirrors? Maybe he was adjusting
something on the touchscreen. People recreating the accident were paying full attention and they were barely stopping in time.

If one is not looking at the road, there is a very high probability a cell phone is not involved so we are playing statistics.

If his wife was with him in the car, we could add in distracted conversation, him getting yelled at for some crap or another to reduce driving focus.

 

[MXWing]

Are we sure that these 7-10 times it tried to kill him, or just a gentle swerve and back? Giving him a false sense of confidence, like a Styrofoam fence painted as steel. And then comes 10.4 and system behaviour changes...

In my view, AP maybe reduces some risks but also introduces new ones.

I also agree to those finding the wheel torque nag to be not good enough, a light grip is not enough. How many seconds from first nag to a warning sound? My previous Citröen tugged the seat belt as a lane departure warning.

Any potentially unsafe behavior for a car with a GVWR of 7,800 lbs travelling at 80MPH one does not take chances on..

If I don't see confidence in the system for a stretch of road, I just don't use it. The very first build that had 80MPH autosteer I was able to take to Vegas from LA and back without any issue. If the X wanted to play Pacman and eat a semi-truck Ghost I would have stopped it immediately. I don't need to try it another 9 times or more.

 

[MXWing]

After Tesla's statement in the public domain, it looks like they are worried about selling more cars and the stock price rather than making the roads safer.

Since when is it Tesla's job to make the ROADS safer. Don't you see you mentioning CALTRANS..

Tesla is not detecting hands on wheel. It detects torque activity on the wheel.

Actually own a Tesla and use AP for a little bit before posting from a false sense of authority.

 

[PhantomX]

With the left tire into the left barrier vertical, the left car beam into the void, and the battery pack into the right vertical, the right beam would miss the barrier entirely. The right barrier vertical would effectively push the front cross bumper bar toward the cabin followed by the front motor carrier, both of which would impart a huge side load onto the main side car beams bending them toward the center line and potentially partially fracturing them at the firewall. Follow that with the impact from the car that hit the X in the #2 lane, and the weakened structure could be removed entirely.

Here is a shot of the undercarriage during impact with a flat obstacle.

View attachment 291720

That's a lot of direction changes in load paths though through different structures. For a 70mph impact, is there enough energy to result in all these structures go past their ultimate strength? And no significant damage to the structure aft of it? There just appears to be a point of discontinuity, and I can't wrap my head around it.

 

[mongo]

That's a lot of direction changes in load paths though through different structures. For a 70mph impact, is there enough energy to result in all these structures go past their ultimate strength? And no significant damage to the structure aft of it? There just appears to be a point of discontinuity, and I can't wrap my head around it.

Yeah, I don't have the FEA skills to figure it out. I looked for crash tests with small posts, but only found side impact. If the impact was folding corrugated profiles(car beams), I can see that ripping/ fracturing the aluminum. The small dimension horizontal width was not made to handle a large side load. I'm thinking the front end is more likely designed to shear off in a side impact to reduce force transfer to the core cabin structure. When the car spun/bounced off the barrier, that may have levered on the beams also.

 

[TEG]

But how does the impact cleanly shear off both beams, the cross braces above in front of dash, and the aft joints of front quarter panels? Both of the beams you pointed out are designed to carry loads in forward and aft direction. If those beams were damaged enough to break away from the car, they would also have transferred enough loads into the main cabin to cause more damages than shown in car. Unless the loads that damaged them were perpendicular to the beams. Same with the cross members and the quarter panels. They are all meant to transfer loads continuously into the rest of unibody structure to distribute impact loads. I keep on asking this because the damage makes no sense as result of head on collision to the barrier unless Tesla really did design front end to break away at specifically energy level.

I think the X may have started to rotate counter clockwise around the barrier once the left wheel hit the edge of the smart cushion.
Perhaps this sheared off the crumple beam due to lateral forces of the rotation?
I go back to something I said way back up thread - it looks like the Model X was "unzipped". I think the smart cushion may have grabbed the left edge and the right edge pulled away around the wall until the windshield, dashboard and much of the frunk area became detached from the rest of the vehicle.

 

[T34ME]

The discussion on this thread has been informative and for the most part, respectful. I appreciate that. Interesting perspectives all around.

I suspect that when the final report comes out there will be plenty of blame/criticism for everyone - CalTrans, driver, and Tesla AP. I have some suggestions for Tesla to make AP safer until FSD is a reality. I'm sure that my suggestions will garner accusations of 'big brother' from some forum members, but the discussion is worth it. Don't give me a 'disagree', tell me why you disagree. A person lost their life probably due to a perfect storm of unfortunate concurrent events. I don't want anyone, particularly a member of this forum, to lose your life even if I have you on 'ignore' because you voted for someone different for President than I did! Here goes:

1. Change the name of the AP system, maybe something like "Tesla Pilot" or "Don't Panic" or 'BFPilot', not that it makes any difference but to make things crystal clear to some people.
2. AP automatically disengages above the posted speed limit, whether it be 25 or 85 mph. AP is not available if speed is greater than posted. Car will still function properly and normally, but you, the driver, are in total manual control.
   
3. Make following distance a function of vehicular speed. The faster you go, the greater the following distance. You can set your following distance greater than the default at a given speed, but you can't set it for less.
4. Require that at least one hand be on the steering wheel at all times, otherwise AP disengages immediately. Perhaps cover the wheel in alcantara so drivers will want to leave their hands on the wheel.
   
5. Both visual and audio warning signals be more robust when AP senses a situation that is potentially dangerous or when it becomes confused.

Any other ideas?

 

[mongo]

The discussion on this thread has been informative and for the most part, respectful. I appreciate that. Interesting perspectives all around.

I suspect that when the final report comes out there will be plenty of blame/criticism for everyone - CalTrans, driver, and Tesla AP. I have some suggestions for Tesla to make AP safer until FSD is a reality. I'm sure that my suggestions will garner accusations of 'big brother' from some forum members, but the discussion is worth it. Don't give me a 'disagree', tell me why you disagree. A person lost their life probably due to a perfect storm of unfortunate concurrent events. I don't want anyone, particularly a member of this forum, to lose your life even if I have you on 'ignore' because you voted for someone different for President than I did! Here goes:

1. Change the name of the AP system, maybe something like "Tesla Pilot" or "Don't Panic" or 'BFPilot', not that it makes any difference but to make things crystal clear to some people.
2. AP automatically disengages above the posted speed limit, whether it be 25 or 85 mph. AP is not available if speed is greater than posted. Car will still function properly and normally, but you, the driver, are in total manual control.
   
3. Make following distance a function of vehicular speed. The faster you go, the greater the following distance. You can set your following distance greater than the default at a given speed, but you can't set it for less.
4. Require that at least one hand be on the steering wheel at all times, otherwise AP disengages immediately. Perhaps cover the wheel in alcantara so drivers will want to leave their hands on the wheel.
   
5. Both visual and audio warning signals be more robust when AP senses a situation that is potentially dangerous or when it becomes confused.

Any other ideas?

My thoughts:
1. Student Driver?
2. Makes system useless given standard driving speeds. Also an issue if the speed zone is mis-recorded or changes. (doesn't it currently have a max delta above posted?) Messing up at 75MPH vs 80MPH doesn't make a lot of difference.
3. Using a braking distance/ speed adjusted following distance is a good idea (Which I thought they did currently)
4. The car can't just stop if you momentary loose contact. (maybe use a different hand on wheel sensor instead of torque)
5. That requires the system to know it is confused or recognize the dangerous situation which is 90% of the problem.

 

[T34ME]

My thoughts:
1. Student Driver?

In my case, Elderly Pilot works too!

2. Makes system useless given standard driving speeds. Also an issue if the speed zone is mis-recorded or changes. (doesn't it currently have a max delta above posted?) Messing up at 75MPH vs 80MPH doesn't make a lot of difference.

"Standard driving speeds" may not be safe driving speeds according to national statistics. But if you do speed and are in accident, you can't blame it on AP from Tesla's perspective. I think a more compelling counter-argument would be, - if one thinks that AP is generally safer than the driver (which I do) then a speed in excess of posted limit would be safer with AP than manually driving. And you got me there! However, I don't think that Elon would want any part in his AP participating in reckless driving ( speeds of 100mph) Yes, there should be a delta for maximum AP speed, but what should that be? 5 over, 10, 15, 20 over? It is subjective.

3. Using a braking distance/ speed adjusted following distance is a good idea (Which I thought they did currently)

I'm not aware of a delta speed adjusted following distance. Anyone else?

4. The car can't just stop if you momentary loose contact. (maybe use a different hand on wheel sensor instead of torque)

The car shouldn't stop, the AP should disengage maybe after 5 seconds with loud warnings.

5. That requires the system to know it is confused or recognize the dangerous situation which is 90% of the problem.

I would think this is a matter for programming as more information is gathered.

Thanks for your perspective and input. Valid points.

 

[bhzmark]

Yes, exactly. And this is very likely why the unreset barrier was so deadly. It is a medieval impaling device, with an apparent lane leading right up to it.

Caltrans really couldn't make it much more deadly and dangerous if they tried.

The guy who designed the waterslide that decapitated a 10 year old faces prison time. The same lethally negligent design and maintenance was committed by Caltrans as well.

Something else I had been pondering about this is if the crumple beams in the front of the X managed to miss hitting the smart cushion. Based on the observed damage, I think it may be possible that the left crumple beam may have gone right into the center of the smart cushion and the right one didn't hit anything. In that case, with a pre-collapsed smart cushion, and Tesla crumple beams possibly missing the smart cushion, there would basically be no slowing of the vehicle at all before impact forces caused massive deceleration forces on the driver.

View attachment 291716
View attachment 291717


-----

I think the government crash tests tend to run vehicles into wider barriers with no holes / channels in them.

View attachment 291718

 

[smac]

@mongo How about this for lateral thinking.

6. Enable AP only if it senses your phone is on charge, or somehow use the app to determine the phone isn't being actively used.


To be honest Apple / Android could both implement something like this for their devices such that when paired to a car hands free kit they go into limited mode aimed at minimising driver distraction, at least until we get to level 3, 4, or 5. It would have meaningful accident reduction even in Level 0 cars.

Here in the UK the penalties for new drivers caught using there phone is an instant ban on first offence, and ban on second offence for those holding a licence for more than two years. Even with such stiff penalties people keep doing it, which is a bit mad when we have the tech to prevent it.

(N.B. I'm not prejudging this accident with this suggestion. Merely adding an additional safety feature.)

 

[MXWing]

The discussion on this thread has been informative and for the most part, respectful. I appreciate that. Interesting perspectives all around.

I suspect that when the final report comes out there will be plenty of blame/criticism for everyone - CalTrans, driver, and Tesla AP. I have some suggestions for Tesla to make AP safer until FSD is a reality. I'm sure that my suggestions will garner accusations of 'big brother' from some forum members, but the discussion is worth it. Don't give me a 'disagree', tell me why you disagree. A person lost their life probably due to a perfect storm of unfortunate concurrent events. I don't want anyone, particularly a member of this forum, to lose your life even if I have you on 'ignore' because you voted for someone different for President than I did! Here goes:

1. Change the name of the AP system, maybe something like "Tesla Pilot" or "Don't Panic" or 'BFPilot', not that it makes any difference but to make things crystal clear to some people.
2. AP automatically disengages above the posted speed limit, whether it be 25 or 85 mph. AP is not available if speed is greater than posted. Car will still function properly and normally, but you, the driver, are in total manual control.
   
3. Make following distance a function of vehicular speed. The faster you go, the greater the following distance. You can set your following distance greater than the default at a given speed, but you can't set it for less.
4. Require that at least one hand be on the steering wheel at all times, otherwise AP disengages immediately. Perhaps cover the wheel in alcantara so drivers will want to leave their hands on the wheel.
   
5. Both visual and audio warning signals be more robust when AP senses a situation that is potentially dangerous or when it becomes confused.

Any other ideas?

1.) No go. Media/Bears/Shorts/Musk and Tesla haters will have a field day. Changing to Tesla ProPilot is not going to work unfortunately. Name is here and they are going to have to double down on it.
2.) Posted speed limit is wrong in certain areas. Posted speed limit also varies depending on hours of the day.
3.) Doesn't make sense to enforce this via AP when you can't enforce it driving manually.
4.) I need to be able to crack open a soda can under certain conditions. AP on the way to the way to Vegas is much different than navigating roundabouts or gore points.
5.) I agree with this.

I applaud your efforts even if I disagree with the majority of your proposed solutions. <3

Wants are unlimited, resources are limited. This is the fundamental theorem of economics. Tesla for sure will get better over time with image recognition to detect more dangerous situations. That's what they need to work on.

What I would love to see is if Tesla can sell a safety retrofit package for older AP cars so they are not reliant on steering wheel torque only.

Internal cabin camera to track eye movement would be a good step.

A Waze type crowd sourcing button to flag potentially dangerous AP areas would also be a good step.

 

[mongo]

@mongo How about this for lateral thinking.

5. Enable AP only if it senses your phone is on charge, or somehow use the app to determine the phone isn't being actively used.


To be honest Apple / Android could both implement something like this for their devices such that when paired to a car hands free kit they go into limited mode aimed at minimising driver distraction, at least until we get to level 3, 4, or 5. It would have meaningful accident reduction even in Level 0 cars.

Here in the UK the penalties for new drivers caught using there phone is an instant ban on first offence, and ban on second offence for those holding a licence for more than two years. Even with such stiff penalties people keep doing it, which is a bit mad when we have the tech to prevent it.

(N.B. I'm not prejudging this accident with this suggestion. Merely adding an additional safety feature.)

These are minimum 5 passenger cars, which phone is the driver's? (even the 3 doesn't help, the owner may not be driving) Sure you could do triangulated cell/ wi-fi position detection, but that is pricey...

 

[PhantomX]

Yeah, I don't have the FEA skills to figure it out. I looked for crash tests with small posts, but only found side impact. If the impact was folding corrugated profiles(car beams), I can see that ripping/ fracturing the aluminum. The small dimension horizontal width was not made to handle a large side load. I'm thinking the front end is more likely designed to shear off in a side impact to reduce force transfer to the core cabin structure. When the car spun/bounced off the barrier, that may have levered on the beams also.

I would love to run an explicit nonlinear analysis to simulate all the scenarios discussed here and understand what really happened. Perhaps Tesla will do it as part of its participation in the investigation.

 

[T34ME]

What I would love to see is if Tesla can sell a safety retrofit package for older AP cars so they are not reliant on steering wheel torque only.

Internal cabin camera to track eye movement would be a good step.

A Waze type crowd sourcing button to flag potentially dangerous AP areas would also be a good step.

Excellent ideas! I think if we all put our heads together, we can come to a consensus of ideas that can make something positive for Tesla out of this unfortunate incident. There but for the grace of dog, go I.