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

A Public Letter to Mr. Musk and Tesla For The Sake Of All Tesla Driver's Safety

This site may earn commission on affiliate links.
... I don't know what exactly is wrong with the benign, generous assumption that the driver was sitting in his seat very shaken up and then he or someone else opened the door; then he heard the beeps and shifted into Park.
...

What is wrong with that assumption is that it is contrary to what he says happened. I agree that we should be generous about what happened immediately after the first post was struck, but we don't have to be forgiving about the actions leading up to that point nor the statements made weeks later. Your "assumption" provides a fairly benign explanation that agrees with what Tesla says the logs show, but if we accept it then that means that Pang is lying about what happened, weeks after it happened.
 
  • Like
Reactions: bhzmark
Some clear things that stuck out as inconsistent:
1) Mr. Pang said Tesla never contacted him to ask about the accident. Tesla says they contacted him the morning after the accident through a translator to check if he was safe and also the next day his wife through a Mandarin speaking employee. This one stuck out to me as either Mr. Pang or Tesla clearly lying.
2) Mr. Pang said there were no warnings from the system prior to impact (although to the police he left an excuse that he didn't understand English in regards to warnings). The logs say there were.
2) Mr. Pang/his friend's account said Autopilot was active through crashing all the posts. The logs said autosteer was disabled before the crash and TACC disabled after the crash.

I'm looking for the "obvious lies" that bhzmark is referring to (bhzmark stated "This isn't a translation issue. And it isn't that his perception is a bit off. He is lying."). Inconsistencies are extremely common after accidents. And you have to realize that Tesla is a multi-billion dollar company with a lot of money at stake that has PR people that get paid to do their job.

Tesla's PR folk are very good. Tesla did NOT say they contacted the driver. Tesla did NOT say there were any audible warnings. Tesla stated that the car DID detect the first impact, and presumably did nothing to stop the car. Yet most people read something completely different.

For #1, Tesla stated "we found a member of the Tesla team fluent in Mandarin and called to follow up. When we were able to make contact with your wife the following day." To me, that sounds like Tesla tried calling the driver, but were unable to reach him. They tried the next day and talked to his wife. Tesla never says they talked to him. That said, it does sound like Tesla certainly *tried* to contact him. But there is no indication that the driver is lying.

For #2, the driver did not say there were no warnings, just that neither he nor the passenger heard any. And sure enough, Tesla appears to have confirmed that. They state that there was a visual warning to keep your hands on the wheel when AP was engaged, and one other warning after that. From what I have read, that would be a visual warning. Again, the driver did not lie.

For #3, the driver said "autopilot continued to drive the car." It is unclear exactly what he means. I doubt he meant that he expected AP to "fix" the situation (drive back onto the road and pretend nothing happened). Tesla says that autosteer was disabled when the driver turned the steering wheel, and adaptive cruise control was disabled "Immediately following detection of the first impact." But that leaves room for [1] the car not detecting the impact right away, [2] the car taking a short time to disable the adaptive cruise control, [3] the car not alerting the driver that adaptive cruise control was disabled, or [4] (most likely) the driver was simply unaware that AP features had been disabled.

So I am not seeing any obvious lies (by the driver or Tesla), as bhzmark says.

If I were the driver, I would ask Tesla to release the log files. It would answer many of these questions. It would likely show that the driver should have done a better job, and that Tesla left out some facts that would have helped clarify what happened.
 
We can continue to split hairs and dissect each other's posts about what happened in Montana (and I realise this is what a lot of people enjoy doing on forums) but it's not going to get us to the truth of what actually happened because a) we weren't there and b) we don't have access to the detailed logs.

But, you've got to admit that Pang and the OP are not doing themselves any favours by avoiding answering the questions and allegations put to them here and it certainly seems as though they haven't gone out of their way to speak with anyone at Tesla, otherwise we'd know about it by now.

Everyone involved has had plenty of time by now to talk to each other, but have chosen not to. Like I said before, that speaks volumes to me.
 
  • Like
Reactions: rscott0
Ah, but how far in advance of the detected collisions was the manual steering input? That makes a big difference, and Tesla does not state. Was it 1 second? Or was it 1/10th of a second? Those are two entirely different scenarios. ...

There is also the possibility that the right wheel dipped off the pavement causing an abrupt steering torque that the logs misinterpret as manual steering input.

I personally am fully willing to believe that the autosteering was operating and drifted (or a hit a pothole that tugged the wheel) into the post. Even so, this is the driver's fault. It does mean that autopilot is imperfect but everybody should already know that. Imperfect does not mean defective. Tesla's don't have run-flat tires; you could strike road debris that cause a blow-out which in turn causes an accident. Should that mean the car is defective, in need of remedies? I don't think so. Autopilot is similar, it handles a lot of situations, but not all situations. You need to remain aware at all times that you are driving with or without AP.
 
  • Like
Reactions: bhzmark and Az_Rael
I hope Naonak doesn't mind, but I took his files, joined them and added some markers to help analyse what happened and I thought it might be useful to everyone to help see what might have happened. (slo-mo of the critical part, too).

Electric Dream, I think your video is really valuable. I had only watched Naonak's videos in full speed. I think I was wrong about the three board barrier. Looking at your slow motion view, the likely trigger appears to be the loss of the yellow side line (or "fog line") together with one of the dashes in the center line while going over the area of new concrete. The loss of the yellow line didn't register as clearly at full speed. There is a short stretch where the only visual lines in the direction of travel are the two edges of the road, so it appears AP tried to center the car in the middle of this one giant "lane" -- right where the cones are.

If this theory is correct, then there is a software algorithm problem that Tesla should fix.
 
  • Like
Reactions: Dangerous Fish
Presumably the manual steering action that turned off autosteer was an abrupt steering action to the right -- into the barriers. He drove himself into the barriers. If it was a manual steering action to the left, then he wouldn't have hit the barriers on the right.
Tesla didn't state which direction the car was steered by the "abrupt steering action". I wish they had. Surely they must know.

We don't know if the driver steered left or right and if that was before or after hitting the first obstacle. It is possible to imagine plausible scenarios for any combination of those possibilities, depending on whether or not the driver was awake or had momentarily fallen asleep (it was after midnight and he had been in the car for many hours, apparently). It is even possible that the passenger, seeing the driver had nodded off, reached over and turned the wheel. There are numerous possibilities. We don't really know what happened, and even if we had full access to the vehicle logs and could understand them that still would not tell us the physiological state of the driver.

It is always advisable to, when speculating about an event about which you have no direct knowledge, to stick to the known facts and acknowledge that there are many unknowns.
 
Electric Dream, I think your video is really valuable. I had only watched Naonak's videos in full speed. I think I was wrong about the three board barrier. Looking at your slow motion view, the likely trigger appears to be the loss of the yellow side line (or "fog line") together with one of the dashes in the center line while going over the area of new concrete. The loss of the yellow line didn't register as clearly at full speed. There is a short stretch where the only visual lines in the direction of travel are the two edges of the road, so it appears AP tried to center the car in the middle of this one giant "lane" -- right where the cones are.

If this theory is correct, then there is a software algorithm problem that Tesla should fix.

Thanks, that is my theory too. Initially, I though the barrier had something to do with it, but when I looked at it more times and frame by frame I didn't think so.

Something that's difficult to know is exactly how far ahead the car's cameras are looking and how fast the information is being processed. There must be a latency involved in all of this which equates to a very small delay between the camera seeing something and the car reacting to it. I don't know if any figures are available for any of that, but it all plays a part in how useful and safe the system can be.

Video would definitely help Tesla improve the algorithms. I do think more videos like this should be submitted. Not necessarily publicly. With better video compression algorithms and cheaper storage, Tesla should be able to record video through the car's cameras and retrieve it along with the data logs they do now. There's a privacy issue which goes along with that of course, but I'd be surprised if they're not thinking about it.
 
We can continue to split hairs and dissect each other's posts about what happened in Montana (and I realise this is what a lot of people enjoy doing on forums) but it's not going to get us to the truth of what actually happened because a) we weren't there and b) we don't have access to the detailed logs.

The problem, for me, is that the driver states that the car while on autopilot "suddenly veered right and crashed into the safety barrier post." If true, that is an an issue that would need to either be explained ("there was something in the road directly in front of the car") or need to be addressed in the software. Such a bug might just affect people on roads AP wasn't designed for, or it could affect driving on more appropriate roads. Tesla did not refute that AP suddenly veered right.

My hunch is that the road curved, AP didn't detect that (which would be understandable if the conditions were not appropriate for AP), and it felt like the car veered when in reality it was the road turning. But why did Tesla not make state that AP made no sudden steering movements? Are they just being cautious (trying not to make it public that AP was simply not advanced enough to sense that impending impact, in conditions AP isn't intended for -- which most people in this forum understand and accept), or are they covering something up (that it did suddenly veer right, where an alert driver may not even be able to avoid an impact).

If the driver chooses not to make the log files available (or cannot reasonably get them), I think Tesla needs to address whether AP did something unexpected.
 
My hunch is that the road curved, AP didn't detect that (which would be understandable if the conditions were not appropriate for AP), and it felt like the car veered when in reality it was the road turning.

I'll risk entering this conversation, where I don't think we have enough facts to come to s solid conclusion. I think the above is likely correct. I've experimented with AP on two lane roads where it shouldn't be used -- with hands on the wheel -- and run it through too sharp a curve. As it keeps going straight I can attest that it feels like it's veering to the right, even though it's not. I haven't let it go off the road before correcting, but I think that if a tire goes off that Montana road, drag from the dirt really would pull it right.
 
the car suddenly veered right

It just veered right in a sudden and crashed into the barrier posts.

why the car veered right suddenly

That is not consistent with:

I've experimented with AP on two lane roads where it shouldn't be used -- with hands on the wheel -- and run it through too sharp a curve. As it keeps going straight I can attest that it feels like it's veering to the right, even though it's not.
 
How so? The only way to get the impression that the car is veering right is to feel like it is veering right. In fact, that is what it means to say 'it veered right'.

Thank you kindly.
Pang states, not once, not twice, but threes times, that the car "veered SUDDENLY" to the right. He could have said the car merely "veered" or he could have said the car "drifted." But he didn't. He said, three different times, that the veering was SUDDEN. He is stating that the AP veered the car suddenly to the right into the wooden posts. He is not to blame, because the car drove into the posts.

Silly Driver posits that driving straight ahead, when the road curves, can give a misperception of veering. But from the google maps, there are no sudden curves, and thus no sudden curving of the road which can give a misperception of veering, or at least certainly not SUDDEN veering.

But those who want to traffic in misplaced charity towards an obvious liar will choose what words of his they want to recognize or not so should conveniently, but politely and kindly, not notice the word "suddenly".

The Pang statement is packed with lies to those who are used to seeing dissembling by people aiming to shift blame. The manufactured outrage. The challenge to "be a man." All marks of lying. That is my evidence-and experience-based infererence.

Also note, stating that he is lying in this case, is not at all a personal attack against Mr. Pang. To Americans, claiming that you are lying is a personal attack, but in other cultures, that have a very different relationship with truth-telling and deception, pointing out lies is not a personal attack, but merely the normal back and forth of conversation -- largely because the opportunity to call out lies is so frequent.
Amazon.com: Lies That Bind: Chinese Truth, Other Truths (9780742554054): Susan D. Blum: Books "Blum points to a propensity for deception in Chinese public interactions in situations where people in the United States would expect truthfulness, yet argues that lying is evaluated within Chinese society by moral standards different from those of Americans. Chinese, for example, might emphasize the consequences of speech, Americans the absolute truthfulness."
 
... an obvious liar ... The Pang statement is packed with lies ... Also note, stating that he is lying in this case, is not at all a personal attack against Mr. Pang. To Americans, claiming that you are lying is a personal attack, but in other cultures, that have a very different relationship with truth-telling and deception, pointing out lies is not a personal attack, but merely the normal back and forth of conversation -- largely because the opportunity to call out lies is so frequent.

You keep calling him a liar. But you have yet to show one of these "obvious lies." I have already refuted several of the statements that others have stated appeared to be lies. I haven't seen you post even one obvious lie.