[UPDATED] 2 die in Tesla crash - NHTSA reports driver seat occupied

[gearchruncher]

There is zero reason why Tesla would deliberately hide the truth at this point.

Like I keep saying, they are just making it worse though, even if they aren't being deliberate. From the Verge article, which is quoting the lead Vehicle engineer:

“All seatbelts post-crash were found to be unbuckled,” Moravy added. Tesla’s advanced driver assist system, Autopilot, can only operate when the seatbelts are buckled.

This is a horrible engineering analysis written to sound definitive to a non-engineer. We have zero idea if the seatbelts were buckled while driving. Even they say that. We also know that the two people in the car were not in the driver's seat when found. So it's highly likely they moved post-crash. So the follow up of "Autopilot, can only operate when the seatbelts are buckled." is completely disconnected from the first part of the thought because we have zero idea of the state of the buckles while the car was driven.

This stands against the background of Elon saying “data logs recovered so far” indicate that Autopilot was not engaged, meaning they have data logs. Those logs would tell you if the seat belts were in use. The fact that Moravy didn't just say "the data logs show the seatbelts were unbuckled while the car was being manually driven" but instead are saying "they were found post-crash to be unbuckled" makes me suspicious that they have not yet 100% told us what they know (and omission is not lying), or the original communication of having data logs indicating AP state was not accurate and the only data they have is based on physical examination of the vehicle.

Can someone please theorize how they are so sure AP was not engaged based on data logs yet only know the post-crash state of the seatbelts?

 

[Maxpilot]

Thus I assume that your point is that these driver assitance modes would not meet any rule similar to the FAA's autopilot guidelines (>3 second safety margin after disengagement to allow operator to take control).

But this is also the case for regular cruise control, which has been available in automobiles for over 60 years. And in those 60 years, no such rule has been enacted by the NHTSA.

Either the rule is not necessary for any vehicle, or if it is necessary then it has to apply to all vehicles, and Tesla would be no different than anyone else. You are cherry-picking Tesla's automation systems specifically for criticism while ignoring the implications of your analysis to other vehicles. This is the very definition of bias.

I don't see bias. It is the truth. FSD does some weird sh!t at the most inopportune time. Driving with FSD is like riding in a car you are teaching your kids to drive. Flying an airplane on autopilot is a lot less stressful.

 

[Dan D.]

This stands against the background of Elon saying “data logs recovered so far” indicate that Autopilot was not engaged, meaning they have data logs. Those logs would tell you if the seat belts were in use. The fact that Moravy didn't just say "the data logs show the seatbelts were unbuckled while the car was being manually driven" but instead are saying "they were found post-crash to be unbuckled" makes me suspicious that they have not yet 100% told us what they know (and omission is not lying), or the original communication of having data logs indicating AP state was not accurate and the only data they have is based on physical examination of the vehicle.

One would presume that data logs on the SD card in the car should match any data logs sent to the Tesla server at the time of the crash. If, for example, the SD data can be recovered and it shows clear discrepancies with the Tesla server data it would indicate data tampering at Tesla and that would be very bad for them. I should think therefore that Tesla would not have tampered with their data but, as you say, they may have selectively released the data. I would like to imagine that when a crash report gets filed by a car, that someone at Tesla would look at it PDQ, or at least before informing Elon. They probably can look at the data fairly quickly, I doubt there is any need to say we have only looked at some of it "so far...". It's likely they already looked at it all (or didn't look at any of it).

 

[MP3Mike]

This stands against the background of Elon saying “data logs recovered so far” indicate that Autopilot was not engaged, meaning they have data logs. Those logs would tell you if the seat belts were in use.

Wow, you know what data is in the logs that they have? How do you know what they contain?

This is a horrible engineering analysis written to sound definitive to a non-engineer. We have zero idea if the seatbelts were buckled while driving. Even they say that. We also know that the two people in the car were not in the driver's seat when found. So it's highly likely they moved post-crash. So the follow up of "Autopilot, can only operate when the seatbelts are buckled." is completely disconnected from the first part of the thought because we have zero idea of the state of the buckles while the car was driven.

It doesn't matter what the state was. We have enough facts:

• Driver's seat belt wasn't buckled post-crash.
• The crash occurred at greater than 30 MPH.

That means:

1. There was a person in the drivers seat. (TACC/AP wouldn't drive without the seat belt buckled.)
2. They were driving manually. (TACC/AP wouldn't have gotten faster than 30 MPH before the crash.)

I suppose the seat could have been empty, with the belt buckled and he was controlling the accelerator pedal with a stick from the back seat. And then after the crash he unbuckled the driver's seat belt before trying to exit the burning car.

 

[gearchruncher]

It's likely they already looked at it all (or didn't look at any of it).

Exactly. Which is why it's weird that they don't just know the seatbelt/weight state already. And actually bad for them if they didn't look at any of it and Elon tweeted "based on data logs recovered..."

 

[gearchruncher]

Wow, you know what data is in the logs that they have? How do you know what they contain?

I don't. I am just shocked if a post-crash datalog contains the AP state, yet not the seatbelt / weight state which controls airbag deployment. What is the point of this immediate datalog to Tesla if not to facilitate emergency responses, where this data could be useful? (and AP is pretty irrelevant)

It doesn't matter what the state was. We have enough facts:

I can 100% agree with this. Which is why Tesla should NOT have brought up seatbelts again. They're the ones that brought up post-crash unbuckled seatbelts on the call. Like I keep saying, they are adding confusion, not reducing it. But they couldn't help themselves because Consumer Reports and others all agree that AP can't work if the seatbelt is unplugged so they just had to try and work that in even though it is irrelevant because they think it's a "told you!" moment.

The statement in the call should have been "All data points to a driver in the seat at the moment of impact with autopilot disengaged. Damage to the steering wheel is consistent with a person present. We look forward to working with the NTSB going forward."

 

[Dan D.]

Exactly. Which is why it's weird that they don't just know the seatbelt/weight state already. And actually bad for them if they didn't look at any of it and Elon tweeted "based on data logs recovered..."

I suppose there could be data corruption in a crashed-car data report. Who knows what a crashed car sends?
Maybe someone can go bash their airbag sensor and then tell us how many GB are uploaded. Thanks.

/s
(no, please don't)

 

[MP3Mike]

I don't. I am just shocked if a post-crash datalog contains the AP state, yet not the seatbelt / weight state which controls airbag deployment. What is the point of this immediate datalog to Tesla if not to facilitate emergency responses, where this data could be useful? (and AP is pretty irrelevant)

My guess is that it wasn't crash related logs that they reviewed. Those probably never made it off the car. (The LTE antenna is in the side view mirror and they often get ripped off during a crash like this before anything would get transmitted.)

But Tesla does keep logs of the number of miles driven on AP, so they may have logs over LTE sent as each mile ticks off and reports if it was AP or manual. No need to include seat belt/weight data. (For one if it is an AP mile you know the seat belt is buckled, and two it isn't relevant to what they are capturing.)

But I am just guessing, as I don't know what Tesla logs or when.

 

[gearchruncher]

But Tesla does keep logs of the number of miles driven on AP, so they may have logs over LTE sent as each mile ticks off and reports if it was AP or manual.

Tesla's own standard is that a crash counts on AP if AP was engaged within the 5 seconds before the crash. Unless they are sending data up every 5 seconds, and a data packet was received within 5 seconds of the impact, they are violating their own definitions by claiming it was not engaged. I really can't imagine they are wasting LTE bandwidth to do that when they can just cache it for when you are on WiFi when they upload all the other AP stuff.

 

[BooMan]

On the quarterly investors call they just said that the steering wheel was deformed meaning that there was a person in the driver's seat. And if I heard correctly all seat belts were unbuckled.

They also spent a week working with NTSB/NHTSA and found that AS would not engage on that street and TACC would only enable while driver was belted at > 5MPH, and even then it would only get up to 30 MPH in the distance covered.

So TACC/AS was not responsible, and a driver was in the seat. (So as suspected probably a failed launch test.)

Does string wheel deformed = driver in driver seat at time of crash? What day the NTSB?

 

[MP3Mike]

they are violating their own definitions by claiming it was not engaged.

How so? This crash isn't included in that data yet. And they probably won't add it until they have all the details. (After all the data is only updated once a quarter, it isn't real time.)

 

[alexgr]

What exactly does "deformed" mean here - bent ? How does that prove there was likely someone in the seat ?

I understood that the deformation was consistent with the driver deforming the steering wheel as a result of this crash.

 

[BooMan]

Wow, you know what data is in the logs that they have? How do you know what they contain?


It doesn't matter what the state was. We have enough facts:

• Driver's seat belt wasn't buckled post-crash.
• The crash occurred at greater than 30 MPH.

That means:

1. There was a person in the drivers seat. (TACC/AP wouldn't drive without the seat belt buckled.)
2. They were driving manually. (TACC/AP wouldn't have gotten faster than 30 MPH before the crash.)

I suppose the seat could have been empty, with the belt buckled and he was controlling the accelerator pedal with a stick from the back seat. And then after the crash he unbuckled the driver's seat belt before trying to exit the burning car.

Not to be a pedant but if we take as fact that the seatbelts were not buckled POST CRASH you can in no way say they were not buckled pre crash. The “fact” that they were unbuckled post crash is rather not telling at all. Just saying. And no not implying AP was involved just saying the post crash seat belt data (however derived) is irrelevant to AP activation

 

[CyberGus]

I understood that the deformation was consistent with the driver deforming the steering wheel as a result of this crash.

Steering wheels are designed to deform, and steering columns to collapse, instead of impaling the driver. Nevertheless, an autopsy should be able to confirm injuries consistent with an impact on the steering wheel.

 

[alexgr]

Steering wheels are designed to deform, and steering columns to collapse, instead of impaling the driver. Nevertheless, an autopsy should be able to confirm injuries consistent with an impact on the steering wheel.

Unless there was another person involved.

 

[Maxpilot]

I'm really surprised that more people posting here are not more concerned with why and how the battery burst into flames. Supposedly.

That should not happen. Instead, I read a thousand posts talking about how a doctor got in the back seat and did a bunch of tricks to get the driving assistance features to drive into a tree. Common sense is commonly uncommon.

 

[stopcrazypp]

Not to be a pedant but if we take as fact that the seatbelts were not buckled POST CRASH you can in no way say they were not buckled pre crash. The “fact” that they were unbuckled post crash is rather not telling at all. Just saying. And no not implying AP was involved just saying the post crash seat belt data (however derived) is irrelevant to AP activation

His point is that the fact the driver's seat belt is unbuckled post crash means it eliminates the most obvious theory (they straight up did the hack that the videos show, with no driver in the seat). It leaves a fairly remote possibility that they did the hack and then decided to unbuckled the seat belt after the crash (why do so if it was the hack?) or it unbuckled itself (extremely unlikely).

 

[stopcrazypp]

I'm really surprised that more people posting here are not more concerned with why and how the battery burst into flames. Supposedly.

That should not happen. Instead, I read a thousand posts talking about how a doctor got in the back seat and did a bunch of tricks to get the driving assistance features to drive into a tree. Common sense is commonly uncommon.

Actually it's common for cars crashing into trees at high speed to catch on fire. Also when Tesla's packs/cells are punctured, they can short circuit and catch on fire. None of this is surprising, nor is it common expectation that the car not catch on fire.

Even when you have "fire-proof" cells like in the Leaf, the car can still catch fire, as there are many flammable materials in typical cars and even if the HV system is completely fireproof, there is still your conventional 12V electrical system and battery, friction/sparks from the crash that can spark a fire.

A Nissan LEAF caught fire in North Texas - cause currently unknown

A spectacular short video posted to Youtube earlier this week shows a Nissan LEAF on fire. According to a post on...

electrek.co

 

[SMAlset]

I would think that with all the testing and refining still going on and with MSM attention to AP/FSD always in the forefront for any crash they cover (for the public’s possible benefit lol), data relating to the state of AP or FSD would be top picks for a quick subset of data to be sent home as soon as the accident happens. Tesla wants to know, the agencies and police responders look for that almost immediately before much else. If its possible to be related to any of those features, seems that’s a trigger for NTSB and NHTSA to send investigators because it is a new emerging technology.

 

[Maxpilot]

Actually it's common for cars crashing into trees at high speed to catch on fire. Also when Tesla's packs/cells are punctured, they can short circuit and catch on fire. None of this is surprising, nor is it common expectation that the car not catch on fire.

Even when you have "fire-proof" cells like in the Leaf, the car can still catch fire, as there are many flammable materials in typical cars and even if the HV system is completely fireproof, there is still your conventional 12V electrical system and battery that can spark a fire.

A Nissan LEAF caught fire in North Texas - cause currently unknown

A spectacular short video posted to Youtube earlier this week shows a Nissan LEAF on fire. According to a post on...

electrek.co

So you accept that? Batteries should not burst into flames. It is not common for Tesla batteries to burst into flames when in a crash, but you say it is? Data does not support that.

In pictures, the front of the car was melted, but the front motor is clearly seen not pushed back into the battery pack. It does not appear that frontal impact caused the battery pack to experience deformation. Someone posted that the storm drain extrusion above the ground was enough to puncture the pack. That storm drain cover shows very little if any indication that it was hit. We need to worry about what someone can do something about (designing batteries so they don't burst into flames) and not what we cannot do anything about (idiots trying to drive from the backseat).