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

[Kincaid]

Those "road cones" weren't even directly in front of the (non-functional) crash cushion. They were practically in the left carpool lane heading toward hwy 85.
I could see someone (or something) being half blinded by the sun in their face, and then noticing those cones and thinking "oh no this lane looks closed!", and swerving into that right "non-lane" with upcoming wall...

View attachment 288911

In the same manner that Autopilot, inattentiveness, and difficult lighting played a role in the May 2016 Model S collision with a truck, those factors created this Model X tragedy.

In the Model S accident, The NTSB unanimously voted to accept that the probable cause of the crash included a combination of factors. The first was the failure of the driver of the truck that the Tesla collided with to yield to the car, and second was the driver’s inattentiveness due to his overreliance on Autopilot.

This Model X accident appears to have been caused by 3 primary factors:

1. The very first factor was the sun at that time of day shining right into the eyes of the driver and the Tesla's cameras and sensors. That lighting factor played into the May 2016 Model S accident as well: the Model S cameras were unable to make a clear distinction between the color of the truck and the background lighting.

2. The glare from the sunlight onto the eyes of the Model X driver and onto his windshield, taken together with the white lines of the roadway on either side of the truncated crash attenuator barrier, made it appear that there was a traffic lane, rather than a section of roadway that led directly to a concrete wall. There was completely inadequate road painting warning and signage warning of a concrete wall directly ahead. In the glare of the sun, the lines painted on the road served to direct drivers directly into the wall.

3. The crash attenuator barrier had not been replaced: there was no buffer to absorb and diminish the Model X. Chances of survival were virtually eliminated.

The lesson to which those of us who drive these cars must steadfastly adhere is this: we absolutely must not trust the ability of the lenses of the cameras and sensors on our vehicles to provide accurate data when challenged by the glare of the sun. We must also never assume that our autopilot software can ever be expected to analyze and defend against improperly designed or degraded highway markers.

I live in Saratoga, CA and have passed by the location of the Model X accident many, many times, often on the way to and from supercharging in Mountain View. That spot, especially when subjected to the glare of the early morning sun, combined with the very poorly designed lane change issues, was an accident waiting to happen.

 

[mongo]

Crazy - looks like a scene in a Michael Bay movie.

So they know about the other 85k times AP was used and 20k YTD, but they don't know about this one at ~9:30am from San Mateo to Mountain View?

The cars are not always sending real time status data. By the time it would have known it failed, it no longer had a computer to send any data (60 MPH = 88 feet per second).

 

[mongo]

I'm still having trouble figuring out where the white car impacted the Tesla that resulted in the damage it had.

My thought is the test landed across the left two lanes pointed to the right. The white car swerves, but caught its left side on the front of the X (near the dash debris), pushing the X toward the concrete.

 

[NeverFollow]

Still looked that way in October 2015.

I notice some improvement between the October 2015 picture and the March 22nd picture:
- A painted Yellow zebra have been added, but still not cushion safety bumper.

 

[22522]

This is pretty poor signage. For comparison, here's what something like that looks like in the UK.

View attachment 289873

Note the clear road markings. This is Spaghetti Junction in the UK and the two left lanes are exiting. They are separated by road markings so people don't change between the two exiting lanes.

Here is Holland. Nice big vertical green chevron sign to alert you:
View attachment 289874
Also a patch of grass which doesn't look like an extra lane. Here's Sweden, the champions of safety:
View attachment 289875
Again, nice big chevron sign, and rumble strips to warn you that you are not supposed to cross the lines.

Really, once you have to remove the sand bags because people crash into them too often perhaps it's time to rethink the signage. I won't even mention having an exit lane on the left of the highway instead of the right, because that happens all the time in the US and just seems crazy to Europeans. The huge trucks are supposed to cut across 7 lanes to exit? OK I mentioned it.

Rumble strips work great. Thank you for mentioning them. They have saved more than a few lives, and might have saved this one.

 

[22522]

The driver was personally known to me. This is a terrible, terrible tragedy and is awful for his family. He leaves behind a wife and two young kids who I'm sure are all in a lot of pain right now.

I am not a Tesla owner, but I found this thread while doing my own research into what could have gone wrong. I've read this entire thread and a lot of people are focusing on the barrier and/or the fire following the crash. Since I learned yesterday that it was Walter involved in the crash, I've been obsessing over the circumstances that could have caused his car to drive straight into the barrier. I'm sure a better barrier could have lessened the damage, but I'm interested in the idea that Autopilot could have contributed to the crash. Knowing Walter, he was not an aggressive or risky driver, and I doubt that this could have happened if he was in full control.

I don't have any more information about the crash than anyone else here. It's been mentioned here before that the driver was an Apple employee. That is correct. I can also confirm that he worked out of Apple's Sunnyvale office, and he was on his way to work at the time of the crash. He should have been continuing straight on 101-S at this junction, so there is no reason he should have been as far left to impact the barrier.

I had a thought/hypothesis yesterday that I would like some opinions on. Not being a Tesla owner personally, I'm not too familiar with exactly how AP works, so you please tell me if this is feasible:

Could it be possible that if AP was engaged, the car was following another vehicle left of the gore point to the 85 ramp? Perhaps the driver realized late that he was in the wrong lane, and signaled the car to change lanes to the right to stay on 101. But, due to the white line markings in the gore point, could the car have moved right into the gore point thinking that it was in a lane? This is total speculation on my part. I'm looking for opinions from those of you who are familiar with the AP technology on the likeliness of this scenario.

Thank you for the knowledge. What you say sounds right. Where I grew up, one would never try to get off a road that rises. But if you did not grow up in a place where bridges go over rivers, and such, that fear might not be programmed in.

I think your scenario makes sense.

 

[cbdream99]

You have probably never traveled on this spot, I timed it and the gore point has enough space for you to travel for 5 full seconds before contact with the wall at 75mph. 5+ seconds! The line condition is not great, but no where close to be bad either.


[QUOTE="cbdream99, post: 2643117, member
If the car already entered the gore area for whatever reasons, then it is possible the car will keep on going until the driver realized that's not a lane and naturally make emergency maneuver.. It's 930am in the morning, I am not sure how bad is the sun's glare or whether it is a contributing factor to the system, but human is easy to miss if the car is already inside the gore area.

I hope Tesla will fully cooperate with the authority and disclose all data they have, and eye witnesses will come out and provide any information.

[/QUOTE]

I actually been to that area before as my HQ is in redwood city, "possible" doesn't mean is a definite, sun glare, may be there is car in front. But I agree it's a conjecture.

 

[verygreen]

They're silent about getting the car's status streaming wirelessly and continuously up to the time of impact when the antenna were destroyed at that time which prevented any further transmission of remote data logging.

Of course there's no such thing as continuous car state reporting to mothership. It does it in bursts every once in a while.

 

[Az_Rael]

I had a thought - do we think this was an AP2 car? The reason I wonder is because the force required to override the wheel in AP2 (at least before 10.4) is a lot higher than with AP1. In my 3 I had a couple of scary disengages at 70mph where I corrected the car with the wheel, AP2 wrestled with me for a bit, then let go, causing me to jerk the wheel. It was scary enough that I stopped using the wheel to disengage because I was worried about losing control at freeway speeds.

There was also this fellow who gave an account of losing control of his AP2 car after using the steering wheel to disengage at 80mph: Scary Experience with AP2.0 Autosteer on the Highway (OP in this thread thought it was a malfunction, but it was probably more likely to be steering overcorrections after the AP disengagement). A couple others in that thread also relayed almost losing control of their cars upon wheel disengagement.



I could see a scenario where AP2 decided to act up right at this bad junction, the X driver disengaged with the wheel, and then due to the high release force, the wheel was jerked at freeway speeds after AP let go. Now the fellow is in a scenario ripe for steering overcorrections and subsequent loss of control of his vehicle.

 

[Owner]

1. The very first factor was the sun at that time of day shining right into the eyes of the driver and the Tesla's cameras and sensors. .

I don't buy the sun glare argument. This was 9:30am -- too far away from sunrise to being a big factor.

I've driven this area a lot (native CA here),

I get sun glare really bad at the end of the day just before sunset near my home and it is both darkish and the sun is in your eyes.

When it is already bright outside at 9:30am, the sun could be a tad annoying but nothing that bad. It is 6:53AM here right now and already quite light outside....

 

[buttershrimp]

I had a thought - do we think this was an AP2 car? The reason I wonder is because the force required to override the wheel in AP2 (at least before 10.4) is a lot higher than with AP1. In my 3 I had a couple of scary disengages at 70mph where I corrected the car with the wheel, AP2 wrestled with me for a bit, then let go, causing me to jerk the wheel. It was scary enough that I stopped using the wheel to disengage because I was worried about losing control at freeway speeds.

There was also this fellow who gave an account of losing control of his AP2 car after using the steering wheel to disengage at 80mph: Scary Experience with AP2.0 Autosteer on the Highway (OP in this thread thought it was a malfunction, but it was probably more likely to be steering overcorrections after the AP disengagement). A couple others in that thread also relayed almost losing control of their cars upon wheel disengagement.



I could see a scenario where AP2 decided to act up right at this bad junction, the X driver disengaged with the wheel, and then due to the high release force, the wheel was jerked at freeway speeds after AP let go. Now the fellow is in a scenario ripe for steering overcorrections and subsequent loss of control of his vehicle.

I set my steering to comfort for this reason, since it’s easier to break auto steer without over correcting, but I realize that it’s a bit sloppy to do this so I’ve become better at quickly flipping the stalk forward before taking over , even when there is an obstacle in the road it’s not a hard habit toward, hope this helps

 

[Reciprocity]

I know that when im in autopilot, the automatic lane change is smart enough to know the shoulder vs a lane. Can someone check if auto lane change is available from the HOV lane into that gore zone?

I agree with @verygreen mostly because he is clearly smarter then me. Tesla might know that Autopilot was engaged at some point, but they wouldnt necessarily know at the moment of the accident. Say if the driver disengaged auto pilot to change lanes into the gore zone.. hence my question above. The only way to change lanes would be to disengage auto pilot if auto lane change is not available.

Looking at everything presented here, I could see this scenario:

1) Driver was in the HOV lane and realized he needed to get over.
2) He disengaged Autopilot and moved into the gore zone looking for a window to merge onto the 101, and maybe even accelerating to merge
3) He was not able to find a window and miss judged the distance to the median, maybe in part because he was not the only car trying to get out of the HOV lane. As others have noticed, this is a somewhat confusing lane and even though this person probably knew this stretch of road very well, its possible that general chaos with other drivers around that point in the road could have contribute to him miss judging the distance.

My guess and this is purely speculation. Autopilot was either engaged and he wasnt paying attention and drove directly into the median. I think this is a much lower likely hood given that this seems to be the persons commute so he would have know the issues at this location and would not have made such a huge mistake of getting into the gore zone and enabling Autopilot, assuming the above that auto lane change would not work here.

I find that one of the two scenarios is the most probable. Its very sad, but being an owner and frequent autopilot user, I to have to force myself to be more disciplined. The worst part is that I am often most relaxed where I am most familiar with the roads and this is actually very bad, because stuff happens all the time and just this stuff that Autopilot is not good at. Its nearly perfect when the situation is very stable but any chaos can cause failure and fast.

 

[timvracer]

2 quick points:

a) Given there is no other damage to the Tesla body indicating another impact that would match the Mazda's damage, I would suspect the Mazda collided with debris (front motor? Frunk/Dash bundle?) rather than the primary Tesla structure?

b) I too believe and speculate that this was an AP failure with an inattentive driver. The road markings are just terrible, and at least 3 times in my usage of AP I have had it do some very unexpected things (jump over a lane) due to confusing markings on the road.

I am not saying AP is "to blame", I have also found AP to be awesome, and super valuable when used properly. In fact, after seeing all the postings of similar intersections in other countries, I am frankly embarrassed for California, who touts itself as some major economic power in the world, has an economic surplus in the budget, yet has 3rd world roads. Just shameful.

 

[tpoltron]

It's aluminum though, so tensile and compressive strengthes are nearly identical. I can see most of the dashboard components being pulled apart if the front did get stuck in the barrier, but I am having a hard time seeing the unibody sustain the damage it did by a tensile load.

Including the welds and other fastening?
I expect the engineers would (rightfully) be designing for extreme loads to be compressive.

 

[zambono]

Just want to add that since a recent update, I have only had one in 2018 my S on AP1 is more likely to try to take an exit from a highway even though I did not indicate a lane change or want to exit. The few times it has happened I have been alert thus just took over, but if the driver of the X wasn't alert or expecting the possibility of the vehicle wanting to exit I can see how this collision happened. Now it could have just been a driver who decided to change lanes at the last moment and didn't calculate too well. I've seen it happen, against the attenuation barrier and it's a hell of a collision at full force.

 

[Zextraterrestrial]

@LMNOP I’ve never found the auto-lane change to be willing to cross a solid white line.

I’ve used it to cross a double-double yellow line into oncoming traffic!! Older software on a ‘17 p90d x but still???wtf!!! I sent in a ‘complaint ‘

 

[Bubslug]

Counterpoint: That scenario does not align with the CHP report that mentioned "driver lost control", unless they only meant that in the sense that no one drives into a barrier on purpose. A straight in impact also seems like it would create a different debris pattern.

From the wreckage the car hit close to center on the barrier, and by the deformation of the battery pack, pretty close to square (I think). Which would agree with the scenario I've described. Also we know the car hit at very high speed by the severity of the damage, so there is no indication of any braking by the car or driver. In addition there appears to be no tire marks showing the trajectory of any emergency maneuvering.

I could be wrong but in any case the safest course of action is to assume a failure of Autopilot, and adjust how you drive your Tesla accordingly until you know different.

 

[mongo]

From the wreckage the car hit close to center on the barrier, and by the deformation of the battery pack, pretty close to square (I think). Which would agree with the scenario I've described. Also we know the car hit at very high speed by the severity of the damage, so there is no indication of any braking by the car or driver. In addition there appears to be no tire marks showing the trajectory of any emergency maneuvering.

I could be wrong but in any case the safest course of action is to assume a failure of Autopilot, and adjust how you drive your Tesla accordingly until you know different.

Good observation on lack of apparent brake marks, does anyone have data on whether the Tesla ABS system induces temporary lockup while panic braking?
It seems to me to have hit centered on the right side of the barrier (based on left tire rim split, left tire rubber on barrier, and dent location on pack).

All drivers (AP or note) should always be aware of and in conscious control of where the vehicle is.

 

[TEG]

Obituary: Former Visceral dev Walter Huang

 

[PhantomX]

Including the welds and other fastening?
I expect the engineers would (rightfully) be designing for extreme loads to be compressive.

Agreed that critical loads would be compression and shear loads in the event of frontal and side impacts. I was just thinking that aluminum is an isotropic material, so the failure strength in tension will be almost the same as compression. Failure mode will definitely be different given the structural design, but I can't quite understand how that would lead to the damage shown on this car.