FSD Beta Videos (and questions for FSD Beta drivers)

[qdeathstar]

Another vid from DirtyTesla

A couple of early interventions while on dirt roads. Car doesn’t perceive overgrown bushes invading road space yet so the driver had to intervene twice to avoid scraping the car against some bushes.

4:25 - car requires intervention on unprotected right turn. Was taking the turn wide with a car approaching which could have resulted in a collision.

7.45 - car starts slowing a bit for no detectable reason, requiring a go pedal intervention. Car then briefly considers an incorrect lane change before correcting itself.

Return trip is sped up because there was no audio due to a glitch. He notes that there was one disengagement (car wasn’t going to change lanes in time to make a turn) and one intervention where he had to hit the go pedal.

All in all, mostly an impressive and incident free vid. Mostly confirms my thought that we might be able to get a wider Beta release by the end of the year, albeit one with user confirmation required for more complex maneuvers like unprotected turns.

edit - 15:20 car hits a thin stick that was sticking up in the dirt road. No damage to car.

I wish we had the side repeater cameras to see how close it really got to the other lane....

 

[qdeathstar]

I guess we’re pretty far from Elon’s fleet of autonomous ride share Teslas.

Obviously, I think the only one seriously debating that is Elon..... but v9 is level 2.

The fact that humans have to intervene is expected.

That said I think they should remove unprotected turns from public release. Other than that I think it could see a much wider release.

 

[rijc99]

Obviously, I think the only one seriously debating that is Elon..... but v9 is level 2.

The fact that humans have to intervene is expected.

That said I think they should remove unprotected turns from public release. Other than that I think it could see a much wider release.

I follow what you’re saying except there are many who tout no accidents with FSD beta but ignore interventions that avoided accidents.

 

[rxlawdude]

I would be concerned of who gets to test these builds on the public roads. I believe in the right hands, the development team can get very valuable data and accelerate advancement that they wouldn't get otherwise. But in the wrong hands, someone will likely get hurt badly or killed. What if someone decided to just push the envelope and decided not to intervene at all? We all know that's not the right way to operate, but it will happen if wrong n safely?

I guess we’re pretty far from Elon’s fleet of autonomous ride share Teslas.

Yep. Except that's not why I bought FSD for our two Teslas.

If someone fails to intervene, in my view of tort law, 100% liability for injuries or death for the driver. Same as a driver with dumb CC that chooses not to intervene.

This ain't rocket science, but I strongly recommend that such morons not purchase nor use FSD.

 

[rxlawdude]

I follow what you’re saying except there are many who tout no accidents with FSD beta but ignore interventions that avoided accidents.

Tautology. Ye Olde "people are sayin'" argument.

Regardless, accidents have NOT happened with driver intervention. You know, with drivers who read and understand they are in command and need to take over in cases where FSD does not behave.

 

[PhantomX]

Tautology. Ye Olde "people are sayin'" argument.

Regardless, accidents have NOT happened with driver intervention. You know, with drivers who read and understand they are in command and need to take over in cases where FSD does not behave.

That's true with FSD Beta. But we know people using Autopilot has been involved in crashes because they ignored warning and over-trust the system. How do we prevent those people from using FSD on public streets when FSD is broadly released? It's one thing they harmed themselves, it's another if they mow down innocent people.

 

[qdeathstar]

I follow what you’re saying except there are many who tout no accidents with FSD beta but ignore interventions that avoided accidents.

Well, honestly I think you are missing the point. No one claimed fsd beta 9 was level 3-5 self driving. Interventions are required for level 2 driving.. I think their point is that it can be released because there have been no accidents meaning that the human expectations/machine actions mesh enough to prevent accidents.

I’m not on board with that though because these drivers represent the safest drivers and definitely not the guy on YouTube recording himself in the back seat.

unfortunately Tesla is at the most dangerous point in developement.

 

[Daniel in SD]

Well, honestly I think you are missing the point. No one claimed fsd beta 9 was level 3-5 self driving. Interventions are required for level 2 driving.. I think their point is that it can be released because there have been no accidents meaning that the human expectations/machine actions mesh enough to prevent accidents.

I’m not on board with that though because these drivers represent the safest drivers and definitely not the guy on YouTube recording himself in the back seat.

unfortunately Tesla is at the most dangerous point in developement.

I think some people feel that it doesn't matter whether or not a wide release of FSD beta would increase road injuries and fatalities.
I'm not sure if they feel the same way about L3-L5 features...

 

[JHCCAZ]

Have to disagree here. At 4:25 you have correctly pointed out the near collision, which 100% would have been a high-speed collision. That alone negates anything positive about the drive. We cannot tolerate this, there is nothing impressive that can make up for such an incident.

Regarding the near-collision at 4:25:
To me this is in the category of maneuvers that are common human behavior but incorrect or improper defensive-driving behavior.

When you are turning into a crossroad with two+ lanes, you simply should not turn into the near lane just because it's currently unoccupied, though there is high-speed traffic in the next lane over.
A) The driver in the next lane over may decide to change lanes at precisely that moment, and may even be looking over his shoulder for blind-spot confirmation, thus not attentive to you suddenly entering the near "clear" lane.
B) The driver in the next lane over cannot know your true intent. Do you actually see him, are you really planning to stay in the near lane and even if so, will you smoothly execute the dangerous squeeze play. If he's a good and defensive driver you're essentially forcing him to brake, swerve a bit over or at least tense up and get ready to do so.

Just a poor move though indeed quite common. FSD should not be doing this.

As I've said, Dirty Tesla is one of my favorites, but with all great respect I say that his comment "the good news is the car did attempt to make the turn" - is not in fact good news, in that situation. (If he simply meant that FSD 9 will now initiate turns without confirmation as a general capability, then OK)

A couple of days ago I had started to write up a message partly about this, but mostly about the unprotected lefts with a stop in the median, that Chuck Cook was bravely attempting. I held back on posting it and then, in a follow-up video, Chuck himself mentioned that it might be more advisable to go right and then U-turn, or otherwise find another route - Bingo!

So although many might disagree with me, I really feel quite strongly that some of these "typical" but inadvisable driving behaviors should be programmed out of FSD. If it makes the owner impatient with the car so be it. I think that any competent driving instructor would say don't drive that way. And we can't let complaint pressure, from people who don't realize that, cause Tesla to attempt maneuvers that are particularly risky and stressful to other experienced drivers. It isn't worth the inevitable accidents, Tesla hatred and setbacks to FSD rollout success.

Finally, aside from all the prior good arguments, it should help the development schedule somewhat to simply rule out these difficult and ultimately unnecessary thread-the-needle maneuvers.

 

[Dan D.]

Regarding the near-collision at 4:25:
To me this is in the category of maneuvers that are common human behavior but incorrect or improper defensive-driving behavior.

When you are turning into a crossroad with two+ lanes, you simply should not turn into the near lane just because it's currently unoccupied, though there is high-speed traffic in the next lane over.
A) The driver in the next lane over may decide to change lanes at precisely that moment, and may even be looking over his shoulder for blind-spot confirmation, thus not attentive to you suddenly entering the near "clear" lane.
B) The driver in the next lane over cannot know your true intent. Do you actually see him, are you really planning to stay in the near lane and even if so, will you smoothly execute the dangerous squeeze play. If he's a good and defensive driver you're essentially forcing him to brake, swerve a bit over or at least tense up and get ready to do so.

Just a poor move though indeed quite common. FSD should not be doing this.

As I've said, Dirty Tesla is one of my favorites, but with all great respect I say that his comment "the good news is the car did attempt to make the turn" - is not in fact good news, in that situation. (If he simply meant that FSD 9 will now initiate turns without confirmation as a general capability, then OK)

A couple of days ago I had started to write up a message partly about this, but mostly about the unprotected lefts with a stop in the median, that Chuck Cook was bravely attempting. I held back on posting it and then, in a follow-up video, Chuck himself mentioned that it might be more advisable to go right and then U-turn, or otherwise find another route - Bingo!

So although many might disagree with me, I really feel quite strongly that some of these "typical" but inadvisable driving behaviors should be programmed out of FSD. If it makes the owner impatient with the car so be it. I think that any competent driving instructor would say don't drive that way. And we can't let complaint pressure, from people who don't realize that, cause Tesla to attempt maneuvers that are particularly risky and stressful to other experienced drivers. It isn't worth the inevitable accidents, Tesla hatred and setbacks to FSD rollout success.

Finally, aside from all the prior good arguments, it should help the development schedule somewhat to simply rule out these difficult and ultimately unnecessary thread-the-needle maneuvers.

My issue with the turn is that it was going wide INTO the far lane and only by Dirty Tesla yanking the wheel did the car not get hit. His car jerks right as he spins the wheel. You have a point about entering the near lane being risky when a car is going past, but this turn went into the far lane (or was going there until the driver pulled it back).

 

[qdeathstar]

I think some people feel that it doesn't matter whether or not a wide release of FSD beta would increase road injuries and fatalities.
I'm not sure if they feel the same way about L3-L5 features...

well, it is a legitimate position, if morally questionable.

Sacrificing lives today for lines tomorrow, maybe....

 

[JHCCAZ]

My issue with the turn is that it was going wide INTO the far lane and only by Dirty Tesla yanking the wheel did the car not get hit. His car jerks right as he spins the wheel. You have a point about entering the near lane being risky when a car is going past, but this turn went into the far lane (or was going there until the driver pulled it back).

Yes I totally understand that point as being the first-order reason for the disengagement, and what came across in the video is that it's also Dirty Tesla's problem with it.

I'm basically saying you shouldn't be doing that in the first place, even less so if you're not going to do it perfectly.

 

[rxlawdude]

well, it is a legitimate position, if morally questionable.

Sacrificing lives today for lines tomorrow, maybe....

Or another false premise based on opinion, and insulting to those who use it properly.

 

[qdeathstar]

Or another false premise based on opinion, and insulting to those who use it properly.

Certainly some people have that view though, similar to how musk said mars would likely be a one way trip at first.

 

[rxlawdude]

Certainly some people have that view though, similar to how musk said mars would likely be a one way trip at first.

I'm not aware that SpaceX Mars vehicles require the passengers to "be prepared to take over."

 

[qdeathstar]

I'm not aware that SpaceX Mars vehicles require the passengers to "be prepared to take over."

Well, there is a pilot

 

[JJRicks]

@JJRicks wondering when you are going to post the compilation

Now that the series is over I can make some compilations, for sure. New schedule means it might take a few weeks though

 

[gearchruncher]

Simply put: speed kills. That shouldn't be difficult to understand.

Just going to flat out ignore the statistics, huh? You want me to "understand" something that is flat out not true?
Yes, you are probably more likely to die if you impact a static object at 70 MPH than 35 MPH, but the statistics make it clear that you are MUCH more likely to hit something on a road that has a 35 MPH speed limit. And as it is always said - speed doesn't kill, speed differentials do.
Let me ask- what speed do you drive? Is it 10 MPH everywhere, because speed kills?
Do you ever ride in an airplane at 500 MPH if speed kills?

Your table got HTML-screwed. Do you mind posting a screenshot as well? I'm interested but my brain is going for a loop with the misaligned headers.

 

[qdeathstar]

Cars kill people 0.o

 

[Mengy]

If this is the performance of your so-called leader of AV that is supposedly 5-10 years ahead. Then we will never solve AV.

That's an oddly derogatory thing to say.