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.
Considering the car is burned pretty severely, I question what could be pulled. Would also be interesting to know is how much data (if any) is transmitted in real time to Tesla.

Anything pulled would have been uploaded right after the crash.

In the event of a crash the car is supposed to upload information.

We don't know if the car had the capability to upload information after the crash, but I think its likely that this functionality survived the initial crash. I doubt anything survived the fire. I don't believe Tesla has a fireproof black box for storing data.
 
  • Informative
Reactions: SO16
But, what happens when you quickly buckle back the seat belt after one gets out of the driver’s seat? Would the car still continue to warn and stop? Or would it go back to driving?

Continues to warn, not sure if it would have come to a complete stop but it seemed like it wanted to. Just tried this in my 2017 S on the highway and had to pull on steering to disengage and take over manually even though foot was on the accelerator. Was able to resume (engage) FSD immediately after unlike when you go too fast and have to restart the car.
 
  • Informative
Reactions: TespaceX
In fairness to The Media, they are reporting the statements made by local officials. It would actually be more unprofessional to report such statements but then add "sounds like bullshit to me", or worse, ignore the story and publish nothing at all.

A publication with integrity would at least get a statement from the auto manufacturer, but Tesla doesn't do that, other than Elon's tweets.
But they actually add their own conclusions and twist to the story. For example, the Drive has the internet search title (changed in the article when you open it) that reads "No one was driving in Tesla autopilot crash that killed two in Texas". Then multiple news portals such as CBS add beef from older (negative) stories about the Autopilot and the fear of the [Autonomous] car. They are NOT just reporting the story, they are twisting it.
 
Here's a link to two young guys in Texas who managed to connect comma.ai to a Tesla S. I'm not suggesting they have any connection to the crash, but Openpilot is open source and there are hints on how others could do it for themselves. Just giving an alternate theory that the victims could have bypassed safety sensors, and made a jury-rigged autonomous system that didn't work that day.


These instructions would also work on an autopilot Model S, but the controls for engaging openpilot vs. the car's autopilot could be confused so caution is advised. To get openpilot working on an autopilot Tesla Model S, omit the section on Bypassing the EPAS kill signal.
 
Last edited:
In AP mode, I have my hand on the wheel, just on the center spoke which keep the AP going without a warning. To my surprise, I was on Ebay looking for towing accessories and noticed several adds for weights you place on the wheel. This allows you to take you hands off the wheel and still satisfy the attention sensor for AP.
That's insane, who the hell would do something like that and why, even, would it be sold. Suicide knobs were outlawed, (in NJ at least), why would these dangerous, stupid devices not be outlaw, hope soon.
 
Seems to me I've read posts in which people mention the car trying to attain its stored highway cruise-control speed upon an inadvertent engagement in a parking lot. And that this is not gradual acceleration, but very strong. I wonder if this could be involved.
To engage TACC or AP, the car must be in drive in the first place. Then, it NEVER accelerates abnormally strong on TACC or AP. And, finally, a touch of the brake pedal deactivates TACC and AP. I guess those stories are mainly from people who don't want to pay for damaged property.

The most obvious scenario is two guys enjoying the awesome acceleration - in an unsuitable location - and losing control. One of the responders has partial knowledge of Tesla and makes a speculative comment at the grisly scene. His co-responders start nodding and the idea takes hold. This balloons into the news story we're all following now, and then it becomes too embarrassing to say "well hold on a minute, we're really not sure just what actually happened here." And none of the so-called journalists want to hear that anyway. So here we are.
I bet there will be no correction stories in major news outlets.
 
Here's a link to two young guys in Texas who managed to connect comma.ai to a Tesla S. I'm not suggesting they have any connection to the crash, but Openpilot is open source and there are hints on how others could do it for themselves. Just giving an alternate theory that the victims could have bypassed safety sensors, and made a jury-rigged autonomous system that didn't work that day.


These instructions would also work on an autopilot Model S, but the controls for engaging openpilot vs. the car's autopilot could be confused so caution is advised. To get openpilot working on an autopilot Tesla Model S, omit the section on Bypassing the EPAS kill signal.

Very unlikely. Comma.ai is sort of an AP1, it works with the front camera pretty much the same way. It needs road markings.

Most likely AP had nothing to do with this crash. Will be interesting to see the "official" conclusion.
 
It is true that AP checks for weight and seatbelt as well as torque on the steering wheel, but these checks can be defeated pretty easily if someone chose to do so.
I just did a little experiment, and AP does not drop if I completely raise my weight off the driver seat. 2021.14.15.

I could climb out of the seat without disconnecting the seat belt. I didn’t try it.
 
  • Like
Reactions: byeLT4 and jsmay311
I agree with many of the other posters here that there are lots of holes and inconsistencies in this story. Lots doesn't add up.

But I just wanted to dispel the oft-repeated idea that AP or TACC could not remain engaged if someone wasn't seated in the driver's seat (or without otherwise defeating the driver's seat occupancy sensor with a weight or whatever).

I'm NOT saying the deceased did this, but if you have the seatbelt buckled behind your back, you can climb out of the drivers seat and into the backseat without AP or TACC disengaging on their own. IOW, the seat occupancy sensor doesn't come into play after AP/TACC is engaged as long as the seatbelt remains buckled.

I just did a little experiment, and AP does not drop if I completely raise my weight off the driver seat. 2021.14.15.

I could climb out of the seat without disconnecting the seat belt. I didn’t try it.
+1. Beat me to it.
 
Last edited:
  • Like
Reactions: iustin
You have to trick the system by putting a weight on seat, and keep the seat belt buckled. At that point, if you want the car to drive for an extended time, you need to put weights on the steering wheel to simulate light torque.

In other words, you have to go out of your way to get AP to stay on without someone in the driver's seat.

My immediate impression is that one of those workarounds eventually failed which turned off AP and it went straight into the tree. If you put a weight on the seat and it's borderline (not enough), it could have disengaged. Same with the steering: if the workaround wasn't enough to keep AP on, it may have disengaged and with the driver being in the back seat, didn't have time to take over. Of course, it could just be a case of AP not being able to negotiate that turn but of all the options, it seems more likely that the kludges you have to do in order to get in the back... may not have worked right (or long enough). Tongue-in-cheek... as if there is a "right" way to do that! :rolleyes:

Mike