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

[DragonWatch]

First, until all the facts are known.

As of the 31 of March 2018, we will have owed our Xena one full year. Anything I say here is strictly my experience and nothing else ~ period.

Within ninety days of owning our car, we had an event that prompted a call to Tesla service. The person we talked to heard the noise/sound, had me troubleshoot some things, and he ran a diagnostic review of our log once the car was turned off and in the garage. He called back a couple of hours later and said we were safe to drive in to the shop on Monday morning. We did not drive the car until Monday morning. I brought ear plugs (did not need them) in the event we had to listen to the sound from Olympia to Seattle, WA. Tesla gave us an equal loner.

First, the loaner had a more up to-date autopilot version than ours. Ours/Xena's was updated a few days later. But I could tell the difference between the loaner and our car. Bottom line, the cause of the initial service call problem was not discovered and they kept the car over a three day period. Thankfully, I can factually state that the trouble has not returned since then either. And, there is nothing on my service documents of the service that indicated any behind the scenes correction took place either.

Second, while using autopilot since it was updated back then (April 2017), I have noticed some issues and learned to adapt to them. Kind of like adaptive software There is a spot along state highway 3 in Washington where the painting of lines was done by a three year old monkey after filling his canteen with moonshine. Up until the (2018.10.4) update about two weeks back, Xena figured playing chicken with on coming traffic at that very location was fun. Therefore, I did not drive with autopilot on that portion of road. We drive Xena about fifteen miles along that stretch of highway 3 at least once a week, and until the recent update we always played chicken; if I wanted to test autopilot. I can now safely state that Xena when under the influence of autopilot, she no longer chooses to play chicken with on coming traffic, and I feel more confident in autopilot. Xena, hell, she will buy off on anything I want to think we did report it, but did not assume there would be a reply.

Third, if you follow me, I know "heaven help you ~ right," my wife and I road tripped down to take the "free" tour of the Fremont factory last September. What I found traffic wise, was that if you were not at a standstill in traffic, you were pushed up to 80+ in some 65 zones. Before you think well he is a speed chicken, please note that I have driven well over 100 mph in Germany only to be passed by a Porsche in my brand new 960 Volvo sedan (one of the first off the line). With the speed control disconnected (I cannot remember the true name) on my Toyota Celica, again in Germany, I ran it well beyond the speedometer and it took minutes to settle back down to read the printed numbers once I took my foot off the gas peddle. My speed control these days is my wife that can also look on the Tesla app to not only see where I am, but how fast Xena is making me drive My take is that there are no speed laws in California. Again before you come at me; I was born (Upland) and raised in California. My wife and I left there permanently when I rejoined the army in 1977. Okay, we returned for four years while I taught ROTC at CSULB so the kids could be near relatives for a bit.

Lastly, before you go off on my dry humor, you must also know that my son at nineteen lost his life as his then girlfriend sped up I-5 towards Canada. She killed two other boys that were in the car that night, but she lived. Our son, his remains were less than desired since the car burst into flames; all we got to see was a white vacuum sealed plastic wrap that did not resemble the boy I kissed goodnight on the forehead as he headed out of the driveway enroute to see his girlfriend and friends the night before. He had just pissed off his sister and I hoped to have conveyed I loved him anyway. By chance we talked to the responding officers and they were extremely remorseful for not somehow responding in a more timely manner that might have saved his life. They had received a call that the car in question was speeding. I think I failed to fully convey to the men that his loss was not on them. I was saddened one more time when the blood test came back that he had been under the influence of alcohol. We did not sue the girl; we figured she would have the burden of those three boys that night with her the rest of her life and a law suit would engrave our son's loss making it worse on us.

I want the truth about what took place in this accident, and I trust Tesla will make it right as best as humanly possible. When the car fires a few years back were brought to light, Tesla did not blame the manufactures of tires, they jumped in, resolved the problem and immediately modified the Model S. To date, Tesla has jumped out ahead of problems. I mention tires, as humor, because back in the early eighties an SUV manufacture blamed car deaths on the tire manufacture. I do not know the result of the lawsuits or if the car tires actually caused the deaths. I do know that a Captain friend of mine died at or about the same time in Graf, Germany because his jeep flipped over during maneuvers. He left behind four children, and wife. Jeeps had a poor center of gravity, and were known for easily flipping.

I believe in responsibility ~ period. Autopilot is new technology, and our social skills are light years behind. I fired a missile (training nuc warhead) at White Sands, New Mexico and it fell short on 21 June 1979:-( Had the investigation not resolved it was a faulty gyro, I would not have been promoted to First Lieutenant. But, I had to wait for the full investigation. The Environmentalists were up in arms since we wiped out a "white" sand dune. My life would have been completely different if the platoon under my command had done something wrong.

I have dealt with computers since 1978, and I can tell you that in that time period, senior officers were scared to death of the new technology. They did not want to be betting their stars on technology. Granted they did not learn their artillery skills the old fashion way with pumpkin tossers, but good old slide rules ruled the day. Hand held calculators were just coming into fashion and allowed on tests. But, we had to learn in depth how to use slide rules. While I have written some computer programs, I know it is not the simple stuff I used to write telling the pc to find drive "D," or moving information from one square to another, or creating a amortization spreadsheet. No, writing code today is mind boggling, and would make my head swim; okay, you probably think my head is swimming too frequently anyway

Let's give it a rest, we are not helping the ones in pain due to their loss, and nothing can happen until the real events around the case are fully investigated.

Following things on a blog are like standing on a bridge watching the stream flow below you and wanting to talk about a single point in time. Not an easy task:-(

Here buy suggestion.

 

[SMAlset]

I agree that the driver is responsible for the collision, but CalTrans' failure to restore the barrier to the proper energy absorbing condition definitely contributed to the severity of his injuries. IMHO, it probably would not have been a fatal collision if the barrier had been restored after the prior collision.

This is new info coming out that I did not hear about. It sounds like Walter would still be alive if the barrier was repaired to the same state as it was before the Prius hit it.

If someone is getting sued here, it's Caltrans.

Whether the suit succeeds or not, I think will come down to where responsibility lies for not fixing a barrier "in case" someone hits it and unfortunately they did. Law is not my area.

Actually this is something we have been discussing since probably Day 1 or 2 and glad to see it being reported more widely now after the Dan Noyes story and Tesla's response. If you go back through the threads from the beginning you'll see TMC members posting video of what they captured on their dashcam from before, after and even that morning 1-1/2 hours before Mr. Huang's accident. Tesla in their blog response showed the crash barrier the day before but our member showed it the morning of given even more credience to the fact we can be pretty certain (well within a number of minutes) that that barrier had not been reset. From news that came out just recently and also mentioned in this thread, there was a previous accident more than a week before this Tesla accident.

A tragic accident. My condolences to the deceased family and friends. Regarding fire, wondering if 100D vs 90D battery pack is more vulnerable to penetration since it appears to have one group in front of the enclosed frame of the main battery pack?

As for the battery pack placement probably the best images we have of the damaged front end of the car and battery pack and protective frame come from Dean C. Smith (video for ABC7NewsBayArea): Dean C. Smith on Twitter If you have an iPad to view it on you can click on the images and enlarge them on your tablet. As you can see the high impact of essentially hitting a concrete wall even cracked and pushed the protective frame. Remember looking at this that there was still a reinforced strong metal bumper guard and other parts of the car that would have been in front of this area that got taken out.

Here's a post TEG made that shows framework with battery installed (I'm assuming this is an X?) to give you more of an idea of what you'd find behind the body. Model X Crash on US-101 (Mountain View, CA)

 

[⚡️ELECTROMAN⚡️]

You can disagree all you want but is the truth, AP might have save you many times but it also tried to kill me more than I would like to. One time didn’t read a line and drove me against the freeway divider.

The truth is you let it crash into the freeway divider. The car is going to crash on AP if you let it.

 

[MXWing]

Actually this is something we have been discussing since probably Day 1 or 2 and glad to see it being reported more widely now after the Dan Noyes story and Tesla's response. If you go back through the threads from the beginning you'll see TMC members posting video of what they captured on their dashcam from before, after and even that morning 1-1/2 hours before Mr. Huang's accident. Tesla in their blog response showed the crash barrier the day before but our member showed it the morning of given even more credience to the fact we can be pretty certain (well within a number of minutes) that that barrier had not been reset. From news that came out just recently and also mentioned in this thread, there was a previous accident more than a week before this Tesla accident.

I apologize for not being as informed as I should have been.

Appreciate the measured and kind reply informing me of such.

For Caltrans not resetting the barrier... does this go up to involuntary manslaughter charges due to criminal negligence?

and WHY THE F***K is NHTSA investigating *T E S L A* and not Caltrans?

 

[AviP]

I would not be surprised at all if the "driver lost control" statement was added by a reporter or news editor. Isn't that what they always say when a car collides with a stationary object?

It's similar to when a plane crash occurs in low visibility, it's referred to as a "controlled flight into terrain".

 

[mongo]

I apologize for not being as informed as I should have been.

Appreciate the measured and kind reply informing me of such.

For Caltrans not resetting the barrier... does this go up to involuntary manslaughter charges due to criminal negligence?

and WHY THE F***K is NHTSA investigating *T E S L A* and not Caltrans?

Last I knew It is NTSB, and the investigation is regarding the post crash fire fighting.
NTSB_Newsroom on Twitter

2 NTSB investigators conducting Field Investigation for fatal March 23, 2018, crash of a Tesla near Mountain View, CA. Unclear if automated control system was active at time of crash. Issues examined include: post-crash fire, steps to make vehicle safe for removal from scene.

8:15 AM - 27 Mar 2018

General note to all: if you put

site:https://teslamotorsclub.com/tmc/threads/model-x-crash-on-us-101-mountain-view-ca.111505

into the google search box along with other key words, search, then click the show similar results it will find most of the previous discussion topics.

 

[kak1154]

All this is further evidence to support he wouldn't have been using AP there (or would have been hyper vigilant), given that he didn't trust it. He was a software engineer (so knew technology) & from what his family reports, did not trust AP in that area.

I think the evidence suggests the opposite. If he says 7/10 times the car veers toward the divider, that suggests he has used AP in that area several times, no? Presumably more than ten.

This tells me that he often (or even always) used AP here and just paid attention and noted when it chose wrong. And especially lately, when AP updates have been coming, maybe he continued to use it, to "trial" the new software, as we are all wont to do. It's not hard to imagine a scenario where he just had an off day, or lost vigilance, had poor visibility, the car did something he wasn't used to, or whatever... leading to the accident.

 

[AviP]

I know everyone is complaining about AutoPilot and making assumptions about what caused the crash. Assuming that AP was to blame, I have a few questions:

First, I'm more curious about the X's front collision detection capabilities. What are the forward scanning capabilities? Why didn't they work? Why was there no evasive action?

Second, does the software lack negative scenario priorities? This would be scenarios which will result in an accident. I would think that every kind of self-driving car software would need directives for negative scenarios. And priorities within those directives. For example, there's 10 people crossing the road and I will hit them if I do not swerve into that telephone pole now!

I don't get the "stay-within-the white-lines" theory and hit the barrier head-on rather than swerve. Could the failure actually be in the front detection. I'm not familiar with the X, so please tell me it's capabilities.

 

[flar]

But based on this info, the front-end damage severity, and the trajectory of the wreckage I strongly believe that the car was not swerving or making avoidance maneuvers before impact.

Note that the design of the track for the barrier pretty much prevents any swerving just prior to impact with the collapsed barrier.

If you hit the collapsed barrier then your wheels are straddling the track for it.
If your wheels are straddling the track for the barrier, then they were straddling it 15-20 feet before you hit the barrier.
If your wheels were straddling the track 15-20 feet before the impact, then you will be hitting it directly head-on.

There could have been some swerving earlier on, but as soon as your front wheels pass the head of the barrier track you will get locked onto a fairly direct head-on trajectory. This directional capture could potentially be intentional because the barrier will be most effective in absorbing energy if it can keep the car locked onto its collapsing profile.

 

[SMAlset]

I apologize for not being as informed as I should have been.

Appreciate the measured and kind reply informing me of such.

For Caltrans not resetting the barrier... does this go up to involuntary manslaughter charges due to criminal negligence?

and WHY THE F***K is NHTSA investigating *T E S L A* and not Caltrans?

Not a problem but getting harder to find past posts to point to given the pages to search through. BTW JoelC shared his dashcam video from the morning of the accident (just 1-1/2 hours apx before the accident) as he was driving into work that morning and you can see the barrier wasn't fixed as of that morning, so even close to the time than the images Tesla put up on their blog.

It's reasonable and expected of them to look into aspects of the car like the data logs, recording modules, battery, AP if that was activated, and any info beyond the car itself that Tesla can help with. I understand from news reports on here that Caltrans is also being looked to during this process.

 

[Tam]

...
I ALWAYS have my hand firmly on the wheel. I use it daily. At least on my regular route, I know where it fails and hold the wheel even more firmly or disengage / re-engage after the place. Call it beta, call it driver assist, etc. I find it useful but I manage it carefully.

That's what I practice also. I too babysit it closely and despite of imperfections, I find it very useful and I love it!

 

[bonnie]

I think the evidence suggests the opposite. If he says 7/10 times the car veers toward the divider, that suggests he has used AP in that area several times, no? Presumably more than ten.

This tells me that he often (or even always) used AP here and just paid attention and noted when it chose wrong. And especially lately, when AP updates have been coming, maybe he continued to use it, to "trial" the new software, as we are all wont to do. It's not hard to imagine a scenario where he just had an off day, or lost vigilance, had poor visibility, the car did something he wasn't used to, or whatever... leading to the accident.

The investigation will be the final word, so not throwing out any opinion.

Mine is based on what I would do & what I think most people with a similar technology background would do. Walter was a software engineer, would not have expected AP to start magically working in that area if he'd had previous problems (and we don't know it was 7/10), and either would not use it or would have been hyper vigilant.

I didn't think there is anyone here on this forum would act differently, but it sounds like you would. So I'm patiently waiting for the accident investigation to conclude.

 

[bonnie]

I would think that every kind of self-driving car software would need directives for negative scenarios. And priorities within those directives. For example, there's 10 people crossing the road and I will hit them if I do not swerve into that telephone pole now!

Autopilot is not self-driving.

 

[mongo]

First, I'm more curious about the X's front collision detection capabilities. What are the forward scanning capabilities? Why didn't they work? Why was there no evasive action?

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

Second, does the software lack negative scenario priorities? This would be scenarios which will result in an accident. I would think that every kind of self-driving car software would need directives for negative scenarios. And priorities within those directives. For example, there's 10 people crossing the road and I will hit them if I do not swerve into that telephone pole now!

I don't get the "stay-within-the white-lines" theory and hit the barrier head-on rather than swerve. Could the failure actually be in the front detection. I'm not familiar with the X, so please tell me it's capabilities.

It is driver assist, not fully self driving. It has no scenario recognition. Stay in lane, if moving car in front slows, slow also.

 

[buttershrimp]

Am I missing something as to the the screen shot of the text? This is a post accident text from a family member reported and shown on news as though it is documentation before the event?

 

[buttershrimp]

Are there any traffic cams in the area?

 

[SMAlset]

Note that the design of the track for the barrier pretty much prevents any swerving just prior to impact with the collapsed barrier.

If you hit the collapsed barrier then your wheels are straddling the track for it.
If your wheels are straddling the track for the barrier, then they were straddling it 15-20 feet before you hit the barrier.
If your wheels were straddling the track 15-20 feet before the impact, then you will be hitting it directly head-on.

There could have been some swerving earlier on, but as soon as your front wheels pass the head of the barrier track you will get locked onto a fairly direct head-on trajectory. This directional capture could potentially be intentional because the barrier will be most effective in absorbing energy if it can keep the car locked onto its collapsing profile.

To illustrate what @flar is saying here's a close up on the barrier track from KTVU's coverage and remember the track generally extends a good distance out from the actual concrete wall to lengthen the travel time of the barrier's movement for more protection in high speed impacts. You can see from the fire fighters foot off the ground there's some good height to the track.

 

[Az_Rael]

Am I missing something as to the the screen shot of the text? This is a post accident text from a family member reported and shown on news as though it is documentation before the event?

It was an account from a family member that Walter had previously complained to Tesla about the AP performance at this intersection. Tesla stated they only found records of a complaint about the navigation. The family states they have the service invoice showing his complaints prior to the accident.

 

[SMAlset]

Are there any traffic cams in the area?

Asked and member answered many pages back, but "No". There's a map showing locations in the bay area but that stretch of 101 has none. Kind of amazing since it is a heavily trafficked area.

 

[flar]

I think the evidence suggests the opposite. If he says 7/10 times the car veers toward the divider, that suggests he has used AP in that area several times, no? Presumably more than ten.

This tells me that he often (or even always) used AP here and just paid attention and noted when it chose wrong. And especially lately, when AP updates have been coming, maybe he continued to use it, to "trial" the new software, as we are all wont to do. It's not hard to imagine a scenario where he just had an off day, or lost vigilance, had poor visibility, the car did something he wasn't used to, or whatever... leading to the accident.

Or, more telling about the state of rush hour driving around here...

Even the slightest stray into that gore point (as part of, say, a reasonably vigilant and supervised attempt to collect data on how often and how far AP might stray) would be considered by the toxically aggressive rush hour drivers in that area as an opportunity to advance at the expense of your escape routes.

You can see this in the video shared earlier in this thread where a white or silver car strays into the gore point as a result of lane and destination confusion and the drivers on both sides of it happily drive right up and box them in (not with intent to harm, but definitely without regard to their plight).