Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

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

This site may earn commission on affiliate links.
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.)
 
They weren't able to recover the SD card data.
Not surprising, but it does make you wonder why Elon tweeted "Data logs recovered so far show Autopilot was not enabled". How did they know that if they were unable to recover data logs? Does a Tesla constantly stream the AP engagement state to Tesla but not other data like seatbelt use or weight in seat sensors?

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.

And then wait- why did they need to do all that stuff with NTSB/NHTSA if we already knew AP wasn't engaged and couldn't have been involved?
 
Last edited:
Trust but verify. Or maybe NTSB/NHTSA wanted to be absolutely sure.
Or Elon made it up that they had a "data log" showing it wasn't enabled.
If you are going to "trust but verify" and think that a log that showed it was not engaged could be lying to you, now you have to question all software and any possible bug, which means a few tests up and down a road don't give you a lot of coverage. You'd want to be testing for any case you could think of that would allow AP to be reported as disengaged when it was engaged, not focusing on how it performs on that road if it somehow managed to be on despite the data showing it off. If you don't trust it logging it's state, why do you trust that it will obey lack of lane lines or acceleration parameters?

I'm in no way suggesting AP was involved here- only that Elon actually didn't know that for sure based on a datalog on April 19th when he tweeted that, and the most logical reason they did all these tests was because they didn't have any data to indicate if it was on or off.
 
Not surprising, but it does make you wonder why Elon tweeted "Data logs recovered so far show Autopilot was not enabled". How did they know that if they were unable to recover data logs? Does a Tesla constantly stream the AP engagement state to Tesla but not other data like seatbelt use or weight in seat sensors?
There are two ways to obtain data logs:
1. Upon a crash Tesla immediately automatically uploads a snapshot of telemetry data from the last few seconds to the server via their data connection -- this is where they got the info that Elon reported that AP was not active.
2. The SD card in the car contains much more data (complete black-box) which would give a lot more information, this is what they could not yet obtain due to the excessive fire damage (but sounded like they still hope some data maybe recoverable yet)
 
  • Informative
Reactions: jsmay311
Teslas intermittently upload data to the Mothership for the purpose of teaching the AI. Perhaps there is always an upload when any of the driver-assist features are activated, which is what prompted Elon's assertion that "Autopilot was not engaged"...
 
1. Upon a crash Tesla immediately automatically uploads a snapshot of telemetry data from the last few seconds to the server via their data connection -- this is where they got the info that Elon reported that AP was not active.
So this data includes AP engaged state for the last 5+ seconds, but does not include seat belt state or weight on seat state? That seems really weird. The weight on seat/seatbelt is tied to airbag stuff so I'd really think this would be a primary bit of data right after a crash. If they knew seatbelt/weight they could have immediately refuted the "no driver" narrative more effectively than "not on autopilot".

Also, this offload requires the 12V system, MCU, and data busses to be functional (offloading via LTE is going to take at least a few seconds), which it is not always after major impacts, so you cannot assume it is always there. There was a fair amount of speculation that this crash was major enough that it was reasonable to assume the 12V system was gone and it was one of the reasons someone could get trapped in the rear seat area and be unable to open a door.
 
Last edited:
  • Like
Reactions: CyberGus
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.)
My takeaway from the investors' call is that they are leaning towards a driver being in the driver's seat and crashing the car into the trees without engaging TACC or Autosteer (due to the damaged steering wheel). Then both driver and passenger unbuckled (or they were never buckled in the first place?).

Though not discussed in the call: Based on this and the accident scene, the driver & passenger were unable to open the doors due to the wedging of the trees. It is possible the passenger was not conscious (or was entrapped) at this point, however the driver must have been conscious as he scrambled into the back seat where he was found.

Either he was unable to open the doors/windows/trunk or was too injured to attempt it.

We know that Tesla has described their batteries as resistant to immediate fire, they are designed to delay fire for as long as possible to allow the occupants the best chance to escape. Either the batteries were too damaged and caught fire quicker, or the occupants were too injured/unable to escape before the batteries finally caught fire. These things may come out in the investigation subject to available evidence. Likely though the fire consumed all of the car before the fire department arrived for the "fire in the trees", because they report they put the fire out in just a few minutes and beyond that they just kept the battery cool for 4 hours. Likely all the damage had been done by that point. The fire must have raged for quite some time before the 911 call, plus the fire department response time and the 3-4 minutes of extinguishing the fire.
 
Last edited:
Either the batteries were too damaged and caught fire quicker, or the occupants were too injured/unable to escape before the batteries finally caught fire
Or the 12V system (completely independent of the HV system) was offline and the rear doors would not open as they require 12V. The 12V battery and fusing is in the front of all Teslas.
 
On the investment call, they mentioned the steering wheel was deformed, so there was likely someone in the driver's seat upon impact. Also, all seatbelts were unbuckled. They weren't able to recover the SD card data.
So this supports the original, most-likely scenario - a failed speed demo. No attempt to circumvent the safety mechanisms. Maybe the two were headed for a an AP demo outside the unincorporated area of the Woodlands, but sadly never made it. The crash would have happened in an ICE Corvette or Hellcat just as easily. Don't expect a correction/retraction from the MSM.
 
Or the 12V system (completely independent of the HV system) was offline and the rear doors would not open as they require 12V. The 12V battery and fusing is in the front of all Teslas.
The NTSB is usually quite thorough in detailing failure modes. If the occupants were unable to self-extricate the NTSB will usually list those problems in the report, and give recommendations to the manufacturer. Being trapped in a crashed car is not unique to the Tesla Model S, but there are always unique differences that can be improved or redesigned.
 
So this supports the original, most-likely scenario - a failed speed demo. No attempt to circumvent the safety mechanisms. Maybe the two were headed for a an AP demo outside the unincorporated area of the Woodlands, but sadly never made it. The crash would have happened in an ICE Corvette or Hellcat just as easily. Don't expect a correction/retraction from the MSM.
The original scenario was that presented by the on-scene First Responders. They said no driver. Subsequent evidence has changed that original scenario.
 
Or the 12V system (completely independent of the HV system) was offline and the rear doors would not open as they require 12V. The 12V battery and fusing is in the front of all Teslas.

No, the rear doors on the S&X can be opened via a manual override even if the 12v system fails. (But most people don't know where it is because they don't bother to read the manual.)