Blog NTSB Says Tesla Was Dropped From Crash Investigation For Releasing Details

Tesla has been removed from the National Transportation Safety Board’s investigation into a fatal accident that occurred while a Model X was using Autopilot, the agency announced today.

“The NTSB took this action because Tesla violated the party agreement by releasing investigative information before it was vetted and confirmed by the NTSB,” the agency said in a statement. “Such releases of incomplete information often lead to speculation and incorrect assumptions about the probable cause of a crash, which does a disservice to the investigative process and the traveling public.”

Still, its unclear if Tesla was removed or voluntarily left the investigation. Tesla said Wednesday that it chose to leave the investigation over the dispute.

“Tesla withdrew from the party agreement with the NTSB because it requires that we not release information about Autopilot to the public, a requirement which we believe fundamentally affects public safety negatively,” the company said in an emailed statement to Bloomberg. “We believe in transparency, so an agreement that prevents public release of information for over a year is unacceptable.”

[Update: 5:30 p.m. ET] Further advancing the public feud, Tesla sent Thursday an updated statement to CNBC saying they left the investigation on Tuesday and plan to complain to Congress.

“It’s been clear in our conversations with the NTSB that they’re more concerned with press headlines than actually promoting safety, the Tesla statement to CNBC said. “Among other things, they repeatedly released partial bits of incomplete information to the media in violation of their own rules, at the same time that they were trying to prevent us from telling all the facts. We don’t believe this is right and we will be making an official complaint to Congress.”

The accident in question took place on March 23rd near Mountain View, Calif. when the driver crashed into a concrete lane divider. Tesla issued a statement after the crash saying that while Autopilot was activated, the vehicle’s recorded data suggests that the crash occurred when the driver failed to put his hands on the wheel after several warnings to do so.

The NTSB said it was “unhappy” with the statement.

“In each of our investigations involving a Tesla vehicle, Tesla has been extremely cooperative on assisting with the vehicle data,” an agency spokesman told the Washington Post. “However, the NTSB is unhappy with the release of investigative information by Tesla.”

Musk then responded to a TMC story posted on Twitter to justify the release of information.

https://twitter.com/x/status/980876926830039041

Following a television appearance by the driver’s family, Tesla again issued a statement saying the  driver ignored warnings to take the wheel. That statement seems to have contributed to the conflict with NTSB.

As noted by Bloomberg, Tesla’s loss of formal status as a party to the NTSB investigation means it can lose access to certain information uncovered in the probe and the ability to shape the official record of the incident.

While rare, the NTSB has revoked party status in other investigations. In 2009, the NTSB revoked the party status of the National Air Traffic Controllers Association in the investigation of a midair collision over the Hudson River. In 2014, the party status of both the Independent Pilots Association and UPS were revoked during the investigation of the crash of UPS Flight 1354 in Birmingham, Ala.

 

[Tam]

...No safety issues there for us to be aware of I am sure or they would have broken the agreement in the name of public safety...

For the fire truck accident, the driver admitted using Autopilot. It's was then documented by the fire department and police.

There was no fatality so there was no urgency.

There was no need for Tesla to reveal the vehicle log urgently.

It could wait 1 or 2 years for NTSB to release it.

It's an easy case to be silent and complying to NTSB's silence rule.

 

[Az_Rael]

For the fire truck accident, the driver admitted using Autopilot. It's was then documented by the fire department and police.

There was no fatality so there was no urgency.

There was no need for Tesla to reveal the vehicle log urgently.

It could wait 1 or 2 years for NTSB to release it.

It's an easy case to be silent and complying to NTSB's silence rule.

So if Walter had survived, the safety issue would have been much less urgent? I disagree.

Just because the fire truck incident didn't result in a fatality (of the Tesla driver or one of the first responders on foot at the scene) doesn't mean there might be some finding from that case that affects the safety of all of us. Actually - that case was really the same root cause that killed Walter - a driver not paying attention with Autopilot engaged and the car hit a stationary object. Tesla should have gone full press after that one in the name of safety and maybe Walter would have been more cautious using AP on his commute.

At my job we consider near misses to be just as serious as actual mishaps.

 

[Tam]

...I disagree...

Like you, I too disagree ( that means I agree with your point.)

However, that is how society weighs the seriousness of an issue even if it's an issue that can cause both no-human-harms as well as human fatality.

If no human is harmed, the issue is considered as non-urgent even though we all know that that same issue can cause fatality next time.

For example, before the first documented Autopilot fatality, there were a lot of people showing how they drove without hands on the wheel.

Even though we all know that that practice can lead to fatality but there was no investigations on those youtubers, there were no urgency.

Why must we wait until there's an accident or fatality in order to classify "no hands on the wheel" problem as an urgency?

I disagree with that and we shouldn't wait but that is the way the world operates.

 

[NCAviator]

Tesla has been removed from the National Transportation Safety Board’s investigation into a fatal accident that occurred while a Model X was using Autopilot, the agency announced today. “The NTSB took this action because Tesla violated the party agreement by releasing investigative information before it was vetted and confirmed by the NTSB,” the agency said...
[WPURI="https://teslamotorsclub.com/blog/2018/04/12/ntsb-says-tesla-was-dropped-from-crash-investigation-for-releasing-details/"]READ FULL ARTICLE[/WPURI]

Here we have the clash of cultures. NTSB was created for the aviation industry. They are well understood there. All the Aviation companies know how they operate. As their investigative mandate has been expanded to other transportation areas by Congress; those cultures are not as use to their rules and process. Rail mostly understands their ways. Autos not so much. Very few auto investigations. Mostly bus and bridges.

There are ways to get information released through the NTSB process. Tesla just did not try to work that process it appears or they tried and the NTSB didn't want to go alone.

Companies can also withdraw from being a "party". The NTSB can also kick you out.

There are major advantages to having the companies part of the investigative process. The public is the greater loss.

NTSB and NHTSA can make it hard on Tesla also. Some of those ways have been discussed here. Tesla has been fighting state governments for so long; they now have a culture of not trying to get along.

We will have to see how that works for them.

 

[Darmie]

Maybe I have missed it somewhere, but where was the story about the vehicle that hit this same barrier prior to this accident? Did NTSB investigate that accident as well?

 

[hacer]

Gosh, in just a matter of 1 day the NTSB has revealed that "A first look showed evidence of metal fatigue where the blade attached to a hub, according to Chairman Robert Sumwalt of NTSB" regarding the Southwest airlines engine failure. They also revealed some details from the flight data recorder such as the altitude and airspeed. These are not safety recommendations, they are facts similar to what Tesla released from their data recorder. I guess the NTSB has a generous double standard. Also Southwest has said a fair amount now about what happened. Is the NTSB going to smack them down too?

 

[jelloslug]

Here we have the clash of cultures. NTSB was created for the aviation industry. They are well understood there. All the Aviation companies know how they operate. As their investigative mandate has been expanded to other transportation areas by Congress; those cultures are not as use to their rules and process. Rail mostly understands their ways. Autos not so much. Very few auto investigations. Mostly bus and bridges.

There are ways to get information released through the NTSB process. Tesla just did not try to work that process it appears or they tried and the NTSB didn't want to go alone.

Companies can also withdraw from being a "party". The NTSB can also kick you out.

There are major advantages to having the companies part of the investigative process. The public is the greater loss.

NTSB and NHTSA can make it hard on Tesla also. Some of those ways have been discussed here. Tesla has been fighting state governments for so long; they now have a culture of not trying to get along.

We will have to see how that works for them.

If the NTSB and the NHTSA try and "make it hard" for Tesla, you can bet that Tesla will go after that.

 

[mongo]

If the NTSB and the NHTSA try and "make it hard" for Tesla, you can bet that Tesla will go after that.

I'm thinking Tesla has better things to do with its time (unless provoked).

 

[alcibiades]

Gosh, in just a matter of 1 day the NTSB has revealed that "A first look showed evidence of metal fatigue where the blade attached to a hub, according to Chairman Robert Sumwalt of NTSB" regarding the Southwest airlines engine failure. They also revealed some details from the flight data recorder such as the altitude and airspeed. These are not safety recommendations, they are facts similar to what Tesla released from their data recorder. I guess the NTSB has a generous double standard. Also Southwest has said a fair amount now about what happened. Is the NTSB going to smack them down too?

You're just setting us up for a game of "Spot the differences".

 

[Chocomonsters]

I doubt that will happen. AP, though still imperfect, is better than most other systems on the road. What can they regulate? You can have cruise and lane departure warning, but you can't actually stop the car from hitting or departing? Stopping it from handling the cases it can detect reduces safety.

They can certainly limit how Tesla advertise AP.

Tesla webpage fails to disclose that its AP is “beta” and that video of FSD is very misleading.

 

[Chocomonsters]

Just how many auto crashes does the NTSB investigate? Over the years my wife and I have been involved in a few, all of which were judged to not be our fault. Fortunately no one died, and no serious injury. None of these were investigated by NTSB. Only two of them had law enforcement make a report. Even the uninsured motorist who fled the scene on foot was never tracked down (suspected illegal alien). Is the NTSB investigating because a crash with a TESLA is so rare? Is it because every TESLA auto crash is big news when every other auto manufacturer builds autos that have multiple crashes every day? Something to think about.

Think NTSB is interested in this crash since it involved AP with fatality. NTSB is involved in Uber’s crash in AZ also since there was fatality and self driving issue.

AP, FSD, automous driving, and etc are new thus under closer scrutiny.

 

[Electroman]

They can certainly limit how Tesla advertise AP.

They should. I hope they tell Tesla to change the name AutoPilot to something else.

 

[MrAustraliaTax]

they are facts similar to what Tesla released from their data recorder

Tesla didn't release only facts. They very deliberately tried to spin the narrative in their favour with specific wording designed to misdirect the reader into blaming the driver prematurely and deflect attention away from the fact that AP was ON at the time of the accident.

 

[mongo]

Tesla didn't release only facts. They very deliberately tried to spin the narrative in their favour with specific wording designed to misdirect the reader into blaming the driver prematurely and deflect attention away from the fact that AP was ON at the time of the accident.

Say what now?
They said AP was on.
They said there were 5 seconds / 150 meters of clear view of the barrier.
They said there was no record of driver interaction fir 6 seconds before impact.
Where is this spin? AP is a driver assistance feature, thus the driver is responsible. Just like if a teenage student driver crashes, the adult in the car is responsible.

In the moments before the collision, which occurred at 9:27 a.m. on Friday, March 23rd, Autopilot was engaged with the adaptive cruise control follow-distance set to minimum. The driver had received several visual and one audible hands-on warning earlier in the drive and the driver’s hands were not detected on the wheel for six seconds prior to the collision. The driver had about five seconds and 150 meters of unobstructed view of the concrete divider with the crushed crash attenuator, but the vehicle logs show that no action was taken.

 

[Az_Rael]

Say what now?
They said AP was on.
They said there were 5 seconds / 150 meters of clear view of the barrier.
They said there was no record of driver interaction fir 6 seconds before impact.
Where is this spin? AP is a driver assistance feature, thus the driver is responsible. Just like if a teenage student driver crashes, the adult in the car is responsible.

They also mentioned the follow distance was set to 1, but how is that relevant to the case if he had 150 meters of clear view? They also said at some point previously in the drive he got wheel nags, but didn’t say when (could have been 20 minutes earlier for all we know - and he did interact with the wheel 6 seconds prior, apparently). I feel the only reason they stated those facts was to paint a particular picture of the driver for the media.

They did not mention at all how AP may have mistaken the lines, if AEB fired off, or if the car gave him a collision warning. Those last three things probably aren’t in Tesla’s favor, even though they would be useful for us to know from a public safety standpoint (their supposed only reason for talking).

And it worked. All the headlines then said he had gotten wheel nags and set the follow distance to minimum, even though those two facts may not be related to the crash at all. No article even mentioned the fact that Teslas have AEB and FCW and why those might not have worked.

 

[mongo]

They also mentioned the follow distance was set to 1, but how is that relevant to the case if he had 150 meters of clear view? They also said at some point previously in the drive he got wheel nags, but didn’t say when (could have been 20 minutes earlier for all we know - and he did interact with the wheel 6 seconds prior, apparently). I feel the only reason they stated those facts was to paint a particular picture of the driver for the media.

They did not mention at all how AP may have mistaken the lines, if AEB fired off, or if the car gave him a collision warning. Those last three things probably aren’t in Tesla’s favor, even though they would be useful for us to know from a public safety standpoint (their supposed only reason for talking).

And it worked. All the headlines then said he had gotten wheel nags and set the follow distance to minimum, even though those two facts may not be related to the crash at all. No article even mentioned the fact that Teslas have AEB and FCW and why those might not have worked.

Seeing the world from my engineer POV, I'd say the media are the ones who spun, esp the conversion of "hands not detected" to "hands not on wheel". My thinking is the close following distance exacerbated the issue with the missing section of gore point.

I do see where you are coming from with the reporting of hands not detected, but nothing about whether AEB/FCW did/or did not activate or why AP did what it did. Again, purely my POV: FCW/AEB are not factors in why the crash happened, but do have relevance to the severity. That said, AEB/ FCW are not catch alls, so having a collision where they did not activate just confirms the known.

Saying the car crashed, AP was on, and the driver didn't steer means that AP steered the car into the barrier. (just not directly)

In terms of the crash itself: car trajectory/ mode, driver data available, and driver inputs are the primary factors. If the logged data is similar to a dash cam, then one may be able to get a good understanding of the crash and what factors were in play. If the report of a clear line of sight and no driver input are accurate, (along with CHP report of non-swerving travel), then other data is something to use to prevent future crashes/ look at how AP/FCW/AEB can be improved, but are not the reason for the impact.

General question in terms of discussion/ linguistics for my own benefit in communication:
I am thinking spin is taking one thing and making it sound like another/ different than the reality. If the data released supports what really happened is it still regarded as spin? To me, spin would be blaming an FSD crash on a driver.

Thank you for the great points!

 

[Az_Rael]

General question in terms of discussion/ linguistics for my own benefit in communication:
I am thinking spin is taking one thing and making it sound like another/ different than the reality. If the data released supports what really happened is it still regarded as spin? To me, spin would be blaming an FSD crash on a driver.

I guess I see spin as a spectrum. Sure, it can be full up trying to make something seem like it is not. That isn’t what Tesla did. But it can also be emphasizing one portion of a situation over another to not highlight both sides of an issue. For example, we all engage in “spin” when we write our resumes. We are only emphasizing our good points, none of our weaknesses.

So I think Tesla very carefully crafted that press release to emphasize the driver’s failings, not mentioning the other backup safety systems that may have reduced the severity of the crash but apparently didn’t. The driver was still not attentive. But I think if the collision alert didn’t go off at all, that is a discussion we should be having too.

 

[mongo]

I guess I see spin as a spectrum. Sure, it can be full up trying to make something seem like it is not. That isn’t what Tesla did. But it can also be emphasizing one portion of a situation over another to not highlight both sides of an issue. For example, we all engage in “spin” when we write our resumes. We are only emphasizing our good points, none of our weaknesses.

So I think Tesla very carefully crafted that press release to emphasize the driver’s failings, not mentioning the other backup safety systems that may have reduced the severity of the crash but apparently didn’t. The driver was still not attentive. But I think if the collision alert didn’t go off at all, that is a discussion we should be having too.

Thanks! That helps me a bunch.

My spin is that I don't think Tesla crafted the release to emphasize the driver's failings. They are pre-biased with knowing exactly how the system works, so they speak from that point of view. The sub 100% use case safety systems acted in their sub 100% way, the AP line follower did its lane following thing, and the human did what humans do.
Internal discussion:
Was AP on? Yes
Why did it follow the line? because it looked like a line with 85% percent confidence.
Did the system override the driver? No, the system recorded no steering torque commands.
Why didn't FCW/ AEB activate? The barrier radar return was 30% lower than the threshold and the vision output is disabled on current versions.

Honesty as a geek engineer, I'd love to know the intricacies of how the FCW/AEB systems work, but given that the media can't even handle detection vs presence, I'd hate to see what they would do with that information.

I'll spare everyone a young child /dog action analogy.

 

[alcibiades]

The media didn't spin "hands not detected on the wheel" to "hands not on wheel". They took Tesla's duplicitous statement exactly where Tesla wanted them to. Good grief. Tesla could have "from an engineer's POV" said "no torque was detected on the steering wheel in the 6 seconds prior to the collision." But they didn't.

"My spin" and "Pre-bias" are post-rational/post-truth expressions.

 

[mongo]

The media didn't spin "hands not detected on the wheel" to "hands not on wheel". They took Tesla's duplicitous statement exactly where Tesla wanted them to. Good grief. Tesla could have "from an engineer's POV" said "no torque was detected on the steering wheel in the 6 seconds prior to the collision." But they didn't.

"My spin" and "Pre-bias" are post-rational/post-truth expressions.

I was attempting to indicate that I comprehend people, including myself, have known or unknown filters/ biases/ POVs.
And "duplicitous statement" is an example of what, exactly? Who is Tesla trying to deceive?

The driver had received several visual and one audible hands-on warning earlier in the drive and the driver’s hands were not detected on the wheel for six seconds prior to the collision.

How would the phrasing "no torque" make a difference? Non-engineers/ those not familiar with AP will immediate ask, "What does that mean?" where upon the response would be "the driver’s hands were not detected" or the "driver did not turn the wheel". However, the second explanation could be wrong if the sensor itself had failed, or the driver did, but too lightly.