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

[Driver Dave]

After looking at a video of another driver on that road, it is easy to see this scenario, as the crucial line between the lane and barrier is horribly faded.

It becomes easy to see, since the lines are in needing of painting, a lane following assistance system could "jump lines" over to the brighter line, into the barrier "lane" if the driver was not paying attention.

When I'm driving and I see not very well maintained lines like this, my attention level spikes up.

The good rule of thumb is:

It's not magic. It follows lines. No lines, no following.

 

[CarbonFree]

No! Don’t do that please, you don’t even know if that’s what happened. Don’t risk another life.

I don't think the chances of a repro. are very good. I suspect Tesla has an emergency system to silently push out Autopilot fixes to the fleet, independent of normal system updates. They've probably already issued an update that hardcodes a fix for this particular location.

 

[SMAlset]

For all the Sherlock’s here testing:

1.) Have we confirmed Mr. Huang was headed to Apple and which campus?

2.) Traffic conditions and weather conditions and time of day?

3.) AP software version on Mr. Huang’s car?

I see software devs are in full force but QA team is MIA.

1) Yes, confirmed by his brother to ABC7news he was going to work which was at Sunnyvale. No mentions that he was attending a meeting at another campus location.
2) Clear weather-wise, and same as was shown by a fellow member who drove it 1-1/2 hours before him. It happened at 9:27am which is after the bulk of the morning traffic so probably less than earlier in the morning. Car pool lane restrictions end at 9am. A good visual idea of traffic then would be to rewatch the Good Samaritan's phone video taken right after they rescued him but honestly by that point the traffic had backed up and cars were trying to get by in the far two right 101 lanes amid the debris.
3) no idea. Don't think Tesla has said.

 

[CarbonFree]

Glad you were able to recover.

For those new to AP, two years ago AP consistently tried to take exits, etc. It has evolved a lot since then...

Sorry, I should have emphasized the 1.0 aspect. AP is much, much better now.

 

[e-FTW]

So an interesting side note - I read that the fire department had been trained in extinguishing a battery fire but chose not to follow the recommendation because they would have had to close an entire side of the highway by running a fire hose across the highway. Apparently part of the recommended approach is to extinguish the fire with a large volume of water - presumably more than is carried on a truck. So overall it seems that Tesla has provided quite a bit of education to fire departments but there is still a practical learning curve.

Yes, that is a part of this event we have not really discussed. The fire chief said in an interview to ABC 7 (IIRC) that following Tesla's recommendation to “secure an additional supply of water” would have required running 2,000 feet of hose, and closing the highway. Not a great.
He also cited concerns over the environmental impact of all that water runoff contained byproducts of combustion. Fair point, not great either.
As for the fact they had to stand around for hours waiting for the car to be safe to move, that seems to come from the unique situation where they could not follow protocols and cut the first responders loop (it was removed by the impact, along with the rest of the front of the car). They ended up in a situation they were not ready for (is HV power out even though we can't cut the loop?) and needed Tesla's assistance. Turns out the 12-volt battery being gone completely did the same job as cutting that loop, but no one had told them that before last Friday I would assume.

 

[jimmy_d]

Ok, so I was wrong about AP not being active, I suck.

Now we need to modify the theory of what happened. We are now assuming that Walter thought he was in the adjacent HOV Lane? The one that would have taken him to the 101? I’m still confused as to how the car goes straight at that point. With a lead vehicle cutting him off an obscuring Lane lines for a moment? What’s the working theory now?

Hey Butter,

It's a weird case. I've driven through there on AP (both 1 and 2) and it always seemed pretty solid to me. The lane markers are white-on-white through that whole section down there, which is generally not a good thing, but aside from early AP1 I never really noticed any lane holding issues there. I mean - the markings aren't terrible, they're just not great. I've driven in it pouring rain with lots of vehicle spray and straight into the morning sun and on flat cloudy days and it's never been an issue. It must be just about the most heavily travelled AP route in the world, so you'd expect them to have at least this one working.

I haven't tried it on 2018.10.4 yet. Here's a photo from a 'bad' day a couple of weeks ago at about 10:45am. It's a bit before the section where the accident occurred (maybe a mile or two?).

So that was on 2018.6.1 and the lanes look pretty terrible under those conditions but the lane holding was really good. This photo is at low speed but traffic was intermittently going 75 and it was no problem. And on a good day the marking are easy to see (it's the rain that ruins them). So it seems odd to me that lane marking on the day of the accident, which was clear, were bad enough to be a big contribution to the problem. Of course glare could be a lot worse at 9:30am but I've driven it in full on glare and AP hasn't been particularly wonky.

Because I've seen people trying to swap lanes there at the last minute I initially assumed that the driver tried one of those maneuvers and got tripped up somehow. Traffic at that location at 9:30 am is a nasty combination of fast and dense and everyone is in a rush at that time of day so if you were to try to swap lanes there at the last minute you could find your target lane closed off before you got into it. As an Apple engineer on his morning commute he probably wanted to go left (towards Apple) but his car ended up on the right side of the barrier. Initially I thought that meant that he was in a different lane from the one he wanted to be in and was trying to change. But that must all be wrong since he was on AP and didn't make any steering inputs in the seconds prior to the accident.

There's the theory that AP incorrectly locked onto the wrong side of the lane split and ended up driving straight down the split. It nicely explains how the car would hit the barrier head on but slightly to the right since the barrier is towards the left of the split zone. But I've never, personally, seen that flavor of screw up and I've seen so many screwups that I sort of feel like I've seen them all - many times. Every time I've seen a lane split failure it was because AP changed it's mind about which side it should be in after the start of the split. It doesn't end up riding the split it just changes over to the other side. This particular split is really, really long so that kind of fail wouldn't have caused an accident here, but being so long it might look more like a lane once you get down it a ways. But if AP isn't fooled at the start then being fooled later and moving over is going to be a lane change. This isn't a curve - it's dead straight there. I've never seen failure to hold a lane on a dead straight road with marked lanes.

But maybe 2018.10.4 has a new failure mode that I haven't seen yet. If I still have 2018.10.4 the next time I drive through there I'm definitely going to try it on that exact path. Another thing that is weird about this theory is that, if you're paying the least bit of attention at all, you'd notice that all the cars around you are starting to go one way and your car is going the other way. Peripheral vision reflexes make it hard to ignore that phenomenon. I wish I could say that the split zone feels different there but so many people drive across it that it's just as clean and smooth as the road. And of course all of the bots dots are worn off and there's no road boundary vibration imprint on the pavement either.

I don't drive with the distance set to 1 so there might be some behavior that happens with close following that I wouldn't have seen. (I never use anything less than 4 and usually stick with 7.) The FOV of the main and narrow front cameras is not all that wide and it could be occluded by a vehicle ahead if it is both large and close, so maybe being on 1 could lead to a situation where the lane markings were blocked by the car ahead? Traffic often slows at that location, so maybe the following distance between the X and the vehicle ahead would get even closer, but that's not consistent with hitting the barrier at 70mph. It's also not consistent with the claim that the vehicle had a clear view of the barrier for several seconds.

I have a hard time seeing how a driver who's familiar with the route and the vehicle and paying even a modicum of attention is not going to notice something weird going on for several seconds in that location. It's not the kind of casual drive where your attention wanders because nothing is going on. The vehicle is going to be accelerating and braking to deal with traffic and would have just finished a gradual turn to the left. It seems like the wrong time of day to be falling asleep, and the wrong situation to be intoxicated. Heart attack? Stroke? That seems like too big a coincidence when you combine it with an AP failure. Everything seems unlikely. But then accidents are rare so maybe the unlikely is to be expected.

What seems to be left is something going on in the car that took the drivers complete attention for many seconds and which was engrossing enough that he didn't notice the visual or audio warnings.

I look forward to seeing the results after the investigation is completed.

 

[e-FTW]

As an Apple engineer on his morning commute he probably wanted to go left (towards Apple) but his car ended up on the right side of the barrier.

Actually, he was going to another large Apple campus, in Sunnyvale. Most likely he was continuing straight on 101 south, and not taking the 85.

 

[jimmy_d]

They have gone through multiple different kinds of crash barriers at that location over the years.
The marks you saw were from the circa 2011 original barrier that was a sequence of wooden posts with a guardrail. It got replaced by sand/water barrels in 2012/2013, then replaced with the "smart cushion" device we have now.

2011 version of the barrier:
View attachment 290809

You can see that people ran into that one too...

There were never any plastic poles there.

I stand corrected. Was going to mention the barrels but thought I must be mis-remembering since there doesn't really seem to be enough room for them there. But I guess I got it backwards.

 

[bonnie]

Hey Butter,

It's a weird case. I've driven through there on AP (both 1 and 2) and it always seemed pretty solid to me. The lane markers are white-on-white through that whole section down there, which is generally not a good thing, but aside from early AP1 I never really noticed any lane holding issues there. I mean - the markings aren't terrible, they're just not great. I've driven in it pouring rain with lots of vehicle spray and straight into the morning sun and on flat cloudy days and it's never been an issue. It must be just about the most heavily travelled AP route in the world, so you'd expect them to have at least this one working.

I haven't tried it on 2018.10.4 yet. Here's a photo from a 'bad' day a couple of weeks ago at about 10:45am. It's a bit before the section where the accident occurred (maybe a mile or two?).

View attachment 290810

So that was on 2018.6.1 and the lanes look pretty terrible under those conditions but the lane holding was really good. This photo is at low speed but traffic was intermittently going 75 and it was no problem. And on a good day the marking are easy to see (it's the rain that ruins them). So it seems odd to me that lane marking on the day of the accident, which was clear, were bad enough to be a big contribution to the problem. Of course glare could be a lot worse at 9:30am but I've driven it in full on glare and AP hasn't been particularly wonky.

Because I've seen people trying to swap lanes there at the last minute I initially assumed that the driver tried one of those maneuvers and got tripped up somehow. Traffic at that location at 9:30 am is a nasty combination of fast and dense and everyone is in a rush at that time of day so if you were to try to swap lanes there at the last minute you could find your target lane closed off before you got into it. As an Apple engineer on his morning commute he probably wanted to go left (towards Apple) but his car ended up on the right side of the barrier. Initially I thought that meant that he was in a different lane from the one he wanted to be in and was trying to change. But that must all be wrong since he was on AP and didn't make any steering inputs in the seconds prior to the accident.

There's the theory that AP incorrectly locked onto the wrong side of the lane split and ended up driving straight down the split. It nicely explains how the car would hit the barrier head on but slightly to the right since the barrier is towards the left of the split zone. But I've never, personally, seen that flavor of screw up and I've seen so many screwups that I sort of feel like I've seen them all - many times. Every time I've seen a lane split failure it was because AP changed it's mind about which side it should be in after the start of the split. It doesn't end up riding the split it just changes over to the other side. This particular split is really, really long so that kind of fail wouldn't have caused an accident here, but being so long it might look more like a lane once you get down it a ways. But if AP isn't fooled at the start then being fooled later and moving over is going to be a lane change. This isn't a curve - it's dead straight there. I've never seen failure to hold a lane on a dead straight road with marked lanes.

But maybe 2018.10.4 has a new failure mode that I haven't seen yet. If I still have 2018.10.4 the next time I drive through there I'm definitely going to try it on that exact path. Another thing that is weird about this theory is that, if you're paying the least bit of attention at all, you'd notice that all the cars around you are starting to go one way and your car is going the other way. Peripheral vision reflexes make it hard to ignore that phenomenon. I wish I could say that the split zone feels different there but so many people drive across it that it's just as clean and smooth as the road. And of course all of the bots dots are worn off and there's no road boundary vibration imprint on the pavement either.

I don't drive with the distance set to 1 so there might be some behavior that happens with close following that I wouldn't have seen. (I never use anything less than 4 and usually stick with 7.) The FOV of the main and narrow front cameras is not all that wide and it could be occluded by a vehicle ahead if it is both large and close, so maybe being on 1 could lead to a situation where the lane markings were blocked by the car ahead? Traffic often slows at that location, so maybe the following distance between the X and the vehicle ahead would get even closer, but that's not consistent with hitting the barrier at 70mph. It's also not consistent with the claim that the vehicle had a clear view of the barrier for several seconds.

I have a hard time seeing how a driver who's familiar with the route and the vehicle and paying even a modicum of attention is not going to notice something weird going on for several seconds in that location. It's not the kind of casual drive where your attention wanders because nothing is going on. The vehicle is going to be accelerating and braking to deal with traffic and would have just finished a gradual turn to the left. It seems like the wrong time of day to be falling asleep, and the wrong situation to be intoxicated. Heart attack? Stroke? That seems like too big a coincidence when you combine it with an AP failure. Everything seems unlikely. But then accidents are rare so maybe the unlikely is to be expected.

What seems to be left is something going on in the car that took the drivers complete attention for many seconds and which was engrossing enough that he didn't notice the visual or audio warnings.

I look forward to seeing the results after the investigation is completed.

Excellent post. And I agree with your conclusion of something taking the driver's complete attention for many seconds - but I cannot understand why. According to family, he'd complained about this exact spot several times. It just doesn't make sense that he'd trust AP to the point of doing other things, in a place where he was telling people it failed. And we may never have an answer. It's tragic, father of young kids, full life ahead.

FYI - there is a GoFundMe for the family to help with expenses, if anyone wants to donate: Click here to support RIP Walter Huang organized by Shawn Price

 

[SMAlset]

Since I know the vast majority of people associate Apple offices with Cupertino, that's not really just the case anymore. They do have numerous building locations there too in addition to Apple Park (the Spaceship) and the Main Campus, but here's a map of their current Sunnyvale locations.

At the top left you will see 101 and North Shoreline Blvd. That's the exit for the Mt. View Superchargers. You'll also notice the flyover off 101 that goes to 85. At the time of the accident he was not that far from his office on Mathilda as you can see.

In fact here's the distance and optimal drive time from the crash area to the office. He was so close to having a normal day. Hope this accident brings more safety improvements to our highway and situational awareness when driving.

 

[zambono]

Excellent post. And I agree with your conclusion of something taking the driver's complete attention for many seconds - but I cannot understand why. According to family, he'd complained about this exact spot several times. It just doesn't make sense that he'd trust AP to the point of doing other things, in a place where he was telling people it failed. And we may never have an answer. It's tragic, father of young kids, full life ahead.

FYI - there is a GoFundMe for the family to help with expenses, if anyone wants to donate: Click here to support RIP Walter Huang organized by Shawn Price

I mentioned before it was confirmed the vehicle was on AP that since the last 2018 update I received, something like 2018.6, the car tries to take some exits. An unfortunate scenario in this case.

 

[buttershrimp]

I am still mystified by this accident. For the reasons @bonnie mentioned and also @jimmy_d , I was almost certain he wasn’t on autopilot because it seemed so outlandish and not it’s typical behavior at all, the only other thing I can think of is that sometimes I can get the car confused during lane changes.... the investigation is key, and this was an AP2.5 car? Video might be recoverable from prior to the accident to see if he was nodding off or looking for a phone. I doubt the veracity of him complaining about 7/10 at that exact location. If this is accurate reporting?

 

[EVie'sDad]

I'm really trying to understand how this could have happened, and it was a shocker to me to hear that AP was active during the crash.

My assumption is that traffic was even heavier at the time of the crash (9:25am on a weekday in Mountain View), so the vehicle involved also would have been detecting a vehicle in front while on autosteer. The speed I'm also assuming would be lower because of the traffic on a weekday, but without definitive information on those items, I can't really get more accurate with my testing.

The speed was much higher given the severity of the impact. In addition, a setting of 1 would wait until the very last second before breaking in an emergency. Not something to trust your life with when barreling down the highway at 65-75 mph.

 

[mongo]

I am still mystified by this accident. For the reasons @bonnie mentioned and also @jimmy_d , I was almost certain he wasn’t on autopilot because it seemed so outlandish and not it’s typical behavior at all, the only other thing I can think of is that sometimes I can get the car confused during lane changes.... the investigation is key, and this was an AP2.5 car? Video might be recoverable from prior to the accident to see if he was nodding off or looking for a phone. I doubt the veracity of him complaining about 7/10 at that exact location. If this is accurate reporting?

Tl;dr; If AP switched from car following to lane following while inside the split, would it concider at least the left line a valid lane marker?

If the 150 meter number from Tesla isn't greatly rounded down, there wasn't a clear view until 50 meters or so past the beginning of the split.
If AP had been tracking a car (reason for not being visible prior) and that car quickly changed lanes, is it feasible that the AP camera looks further down the road and interprets where it currently is as being in a lane?

 

[JTseng]

One missing puzzle in light of the autopilot protecting the occupants in case of imminent collision. If and when the autopilot determines a potential collision and the driver fails to take a corrective action, there should a "final" alternative to save the day. I believe Tesla engineer could learn from this sort of tragedies. Otherwise, the driver will always find himself/herself playing guessing game with autopilot as to when to "hand-off" the responsibility. Clearly, according to the Tesla's statement, the driver failed to take the corrective action for 6 seconds/150 meters. This should be intervened by the autopilot's collision avoidance features. This is the area of autopilot build-in safety and should be taken very seriously.

 

[Xenoilphobe]

Now I feel better about buying my Kia. First the massive recall of over 100K cars and now this. Maybe I'll buy one in a few years when they've had time to reflect and understand how to build these cars safely.

6abc.com/video/embed/?pid=823918

Police identify victims of fiery crash on I-76

Tempe woman says her vehicle unexpectedly burst into flames

Recall summaries for new and used Kia vehicles.

Kia Forte

Kia Motors America is recalling 5,245 2013 Kia Forte vehicles because of a front seatbelt problem. Read More

Kia Motors America is recalling 86,880 2014 Kia Forte sedans because the cooling fan resistor may overheat and melt, leading to an increased risk of a fire. Read More

Kia Optima

Kia is recalling the 2011 Hyundai Sonata, because of rollaway risk and brake lights that may not illuminate. Read More

Kia Sedona

Kia Motors America is recalling 97,884 2006-'12 Kia Sedona minivans in salt-belt states because of a suspension problem. Read More

Kia Motors America is recalling 219,800 2006-'14 Kia Sedona minivans for a hood latch problem. Read More

Kia is recalling 79,867 2006-'12 Kia Sedona minivans because of a corrosion problem in cold-weather states. Read More

Kia Sorento

Kia Motors America is recalling 377,062 2011-'13 Kia Sorento SUVs for a rollaway risk. Read More

Kia Motors America is recalling 12,361 2016 Kia Sorento SUVs to replace the accelerator pedal assembly. Read More

Hyundai is recalling 20,300 2013 Hyundai Santa Fe Sport vehicles and Kia is recalling 9,345 2014 Kia Sorento vehicles because a fractured front axle may increase the rollaway risk. Read More

Kia Soul

Kia Motors America is recalling 208,858 2014-'15 Kia Soul and Kia Soul EV wagons because a section of the accelerator pedal may bend and fracture. Read More

Kia Motors America is recalling 95,314 2010-'13 Kia Soul hatchbacks because a headliner plate may detach during deployment of the side curtain airbag. Read More

Kia is recalling 51,641 2014 Kia Soul hatchbacks because the steering gear assembly may separate. Read More

Kia Soul EV

Kia Motors America is recalling 208,858 2014-'15 Kia Soul and Kia Soul EV wagons because a section of the accelerator pedal may bend and fracture. Read More

Kia Sportage

Kia is recalling 11,961 2014 Kia Sportage crossovers because a label has incorrect tire pressure information printed on it. Read More

 

[Super_Popular]

ABC News just said that the X was in AP and his hands where not on the wheel, even though the car alarmed him to put his hands on the wheel. Wondering what version of AP this X had.

 

[Xenoilphobe]

You do know the recall is about a part on a steering assembly that Bosch makes for Tesla right?

That they're paying for the recall.

Bosch has been in business for a long time, and they make really good stuff.

The steering rack is used by Land Rover and many other vehicles - people are idiots -they don't realize most of the parts are off the shelf and used in a lot of other vehicles. Example - Takata air bags... recalls for this will most likely be a ranger visit to screw in four stainless steel bolts...

 

[Az_Rael]

I marked out on the map ~150 meters to the start of the barrier as shown on satellite:

Here it is to the collapsed location:

 

[iaflyer]

No. The car ahead is in the correct place in the lane. The black line you indicated as "painted line gone" was never a lane marker - it is a crack/seam between two different freeway surfaces. The freeway used to be more narrow here and they expanded it to add lanes. They left behind different textures that almost look like lane markers in the wrong places.

Here is your picture but with different labels:
View attachment 290811

I think you're spot on with your summary. I saw the faded line in other guy's video too. My guess is that the Autopilot latched onto the brighter white line leading left exit, thinking it was the left white line, but unfortunately it was the right white line of the exit.

This accident is an example of why I think purely autonomous vehicles are a lot farther off than some would think. In a perfect world, with standard markings that are brand new (like a test track...) the automation will work great. But get in the real world with faded lines, construction, obstructions to the markings, glare, precipitation... a human brain can figure it out, or slow/stop until they do, but automation has a hard time not following the pre-set rules.