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

FSD Beta 10.69

This site may earn commission on affiliate links.
One more sample test.

This one is well done. Failure per the terms though - slightly more than 5-second gap was missed on far side (usual caveats about potentially not seeing people who changed lanes, but we’ll see that in the full video when we actually see results from within the car).

Barely managed to avoid failure on near side - smallest gap was slightly less than 5 seconds.
....but wouldn't it be too late to incorporate data gathered today in a release that is supposed to happen in a few hours?
 
  • Funny
Reactions: KArnold
....but wouldn't it be too late to incorporate data gathered today in a release that is supposed to happen in a few hours?
Never too late, of course. Not like they cannot keep doing releases. Of course, the terms would apply to whatever is the first version that Chuck tests. That may not be 10.69; we’ll see.

I have to say this looks like it is going to be a nailbiter. We’re going to see failures of course - Elon’s promise will certainly be broken (this is what happens when you say 100%; very silly!). But will the failure happen in the first 10 attempts?
 
There was almost a 12 second gap before another car came in the same lane. 5 seconds if you're counting cars in other lanes...but it should go if the left most lane is clear
Sure….but… We’ve previously discussed this, but I am fine with waiting for two adjacent lanes to be clear (I do not proceed assuming no lane changes will occur). Obviously you could leave a little less margin with someone in the middle lane (they can end up abreast of you when you are up to speed). But for simplicity I am allowing 5 seconds, two lanes. Realistically a good human would go with ~5 seconds in target lane and ~3-4 in the middle lane. I think it is ill-advised to go with less than 2-3 seconds to someone in the middle lane. They could hit you. I mean, if no one else is in the car, I might go, and there is no issue…but I’d be at 60mph in 3 seconds, which is not great for relaxing passengers.

It’s also a failure to cause another driver to prudently alter their behavior (cause them to actively drive defensively), so there’s a balance here. If I were in an oncoming vehicle, I would slow down if someone turned into an adjacent lane with no margin. That’s a failure by the person turning.

In any case this was a failure per the relaxed criteria.
 
Last edited:
Sure….but… We’ve previously discussed this, but I am fine with waiting for two adjacent lanes to be clear (I do not proceed assuming no lane changes will occur). Obviously you could leave a little less margin with someone in the middle lane (they can end up abreast of you when you are up to speed). But for simplicity I am allowing 5 seconds, two lanes. Realistically a good human would go with ~5 seconds in target lane and ~3-4 in the middle lane. I think it is ill-advised to go with less than 2-3 seconds to someone in the middle lane. They could hit you. I mean, if no one else is in the car, I might go, and there is no issue…but I’d be at 60mph in 3 seconds, which is not great for relaxing passengers.

It’s also a failure to cause another driver to prudently alter their behavior (cause them to actively drive defensively), so there’s a balance here. If I were in an oncoming vehicle, I would slow down if someone turned into an adjacent lane with no margin. That’s a failure by the person turning.

In any case this was a failure per the relaxed criteria.

If they generalize the solution to wait for adjacent lanes to be clear, you will continue to have other types of turns including UPRs where the car will keep waiting when there is an opening leading to people honking behind you. Sounds like what you're saying is it would be an improvement...but it absolutely is still not good enough if it always has to wait for adjacent lanes
 
you will continue to have other types of turns including UPRs where the car will keep waiting when there is an opening leading to people honking behind you. Sounds like what you're saying is it would be an improvement...but it absolutely is still not good enough if it always has to wait for adjacent lanes
I always wait for adjacent lanes to clear when I drive (in the appropriate circumstances of course). I don’t get honked at. This has been discussed before anyway. The timing requirements are different but it is fine to wait for clear lanes to avoid any chance of a collision.

With three lanes of traffic, depending on speed, on a right turn, I may ALSO wait for all three lanes to clear, or at least for traffic to pass the obvious “danger” point.

It’s just a math and trajectory problem.

In this case, to simplify the bet, it is:

5 seconds, 5 seconds, 6 seconds, 5 seconds, 5 seconds, no gap requirement (should not gate go/no-go) (from near lane to far lane).

In reality it is more like 5,5,6,5,3,x

For a right turn it would be something like 4,3,x
 
Last edited:
Testing in the rain. Nice!
0/1. (For the tweet - there is a longer video, will score later.)


Intervention or not, that is a failure, since definitely the human driver should intervene in that case, and that is not normal driving behavior. I suspect that was a little accelerator kick from the driver to get it to the right spot, but failure either way.
 
For the tweet - there is a longer video, will score later.)
Full video linked above (and below!) is 4/8. Disappointing. Hopefully they can improve it. It’s weird and hard to understand why it waits sometimes. This is not the 100% Elon promised (though it is better than the 30% I expected a month ago, I am a Negative Nancy).

The first attempt was a near collision! Totally unacceptable. I just don’t see why they wait for such huge gaps sometimes and still end up in this position - really need a lot more assertiveness after making the go decision. There was plenty of time; the blue car arrived 8 seconds after the window opened.

If we make the assumption that they can see, this seems like an easy thing to fix. Just GO.


98725DFF-ADC4-4F6D-B997-CFA6B65DAA14.jpeg
95CAC403-CBE2-4120-9A7D-BDA938E2134A.jpeg
 
Last edited:
  • Like
Reactions: LowlyOilBurner
Lol who thinks they will actually release on 8/20? Highly doubtful based on how elon misses all his timeline predictions
If they want the high success level on this turn (which was promised), they are not going to be able to release. Unless there is a “go faster” knob and a “proceed more quickly to pose” knob they can simply turn. Looks like lots of work to do still.

Of course they could just not fix this turn, and proceed. It’s not like meeting Elon’s 100% success metric matters. Depends on how good they think it is and what other problems and regressions they are dealing with, I guess.
 
What a terrible waste of time....
What is? It’s very important from a demonstration-of-capability standpoint to be able to safely do this every time. It’s an easy left turn but with substantial requirements on perception. If it can’t do this I am not sure what is the point (since this would also likely mean it could not turn right reliably, either, or perform merges into traffic from a stationary position, or correctly position itself on the road, or see reliably over 200m!)


It’s definitely a waste of my time, though, lol.
 
Last edited:
Full video linked above (and below!) is 4/8. Disappointing. Hopefully they can improve it. It’s weird and hard to understand why it waits sometimes. This is not the 100% Elon promised (though it is better than the 30% I expected a month ago, I am a Negative Nancy).

The first attempt was a near collision! Totally unacceptable. I just don’t see why they wait for such huge gaps sometimes and still end up in this position - really need a lot more assertiveness after making the go decision. There was plenty of time; the blue car arrived 8 seconds after the window opened.

If we make the assumption that they can see, this seems like an easy thing to fix. Just GO.


View attachment 842441View attachment 842442
Just an observation. You can make up you data and numbers but can you please stop saying he “promised” 100%. He has never said “promised“ at any time. He said they were At 90% and getting close to 100%. There is no word “promise” in any of his statements. Just repeating it over and over does not make it fact.
 
Just an observation. You can make up you data and numbers but can you please stop saying he “promised” 100%
Ok, he committed to 100% (if it is called 10.69, if not, it is ok if it doesn’t solve the turn, as I hear it). Seems like a distinction without a difference, but I agree he did not say “promise.” That was just an equivalent paraphrase.

“We’re gonna solve Chuck’s turn…yes, yes…100%…absolutely….you know, we have a lot of respect for valid criticism”


He said they were At 90% and getting close to 100%.
Yes, as I have previously mentioned, he said this afterwards.

You can make up you data and numbers
What am I making up, specifically? Please be specific. I am trying to be as concrete as possible. It is clarifying.