EVNow
Well-Known Member
From Waymo’s perspective Norway might as well be Serangeti.In the end of the day I am just interested that it works where I drive personally, not the plains of Serengeti
-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
FSD rewrite will go out on Oct 20 to limited beta
- Thread starter diplomat33
- Start date
About 10 days before the 3 months mark of @powertoold prediction. I wanted to give an update. The last few videos from Tesla YouTube evangelists showed how the FSD computer crashed partially and fully multiple times in a single drive. This is another problem that fans like @powertoold , @mspisars and @EVNow ignore. The fact these are computers and computers and software are known to crash. Even software that have been validated and tested can crash. We see this everyday with our own computers, laptops, phones, etc... it shows you what Waymo is able to do in Phoenix is nothing short of amazing.
Nevertheless DirtyTesla did a disengagement and intervention tally in his latest video and out of 10 miles he had 6 disengagements and 4 interventions.
That’s a long way from @powertoold confident prediction that they will be at level 5 with a disengagement over 150k miles in 6 months. We have 3 months left and it’s still in the tens. Makes you think weather you fans should keep using Tesla fandom logic to analyze info or actually objective reason and educated logic.
@powertoold and @mspisars said it was game over. That no one could catch Tesla. That no one is even remotely close. Essentially saying Tesla is 5-10 years ahead.
Yet here we are with a disengagement every 10-20 miles (and that’s me being generous) in city environments.
The question then becomes if you put a competitor like Waymo in for example any city that Tesla operates in. Let’s say Detroit. Waymo can even have a 1 month setup time. Map the entire city and setup everything for a month. Will Waymo have a disengagement better than 10-20 miles?
Let your answer to that determine whose ahead. Notice that we do have that exact same scenario going on publically in California and the disengagement report proves that Waymo is easily better than 10-20 miles.
Let’s then take mobileye who doesn’t need any setup time other than maybe a few days (such as in Germany). Will their disengagement be better than 10-20 miles?
There again is your answer to whose ahead.
Educated Logic & Reason, not emotion and fandom.
Nevertheless DirtyTesla did a disengagement and intervention tally in his latest video and out of 10 miles he had 6 disengagements and 4 interventions.
That’s a long way from @powertoold confident prediction that they will be at level 5 with a disengagement over 150k miles in 6 months. We have 3 months left and it’s still in the tens. Makes you think weather you fans should keep using Tesla fandom logic to analyze info or actually objective reason and educated logic.
@powertoold and @mspisars said it was game over. That no one could catch Tesla. That no one is even remotely close. Essentially saying Tesla is 5-10 years ahead.
Yet here we are with a disengagement every 10-20 miles (and that’s me being generous) in city environments.
The question then becomes if you put a competitor like Waymo in for example any city that Tesla operates in. Let’s say Detroit. Waymo can even have a 1 month setup time. Map the entire city and setup everything for a month. Will Waymo have a disengagement better than 10-20 miles?
Let your answer to that determine whose ahead. Notice that we do have that exact same scenario going on publically in California and the disengagement report proves that Waymo is easily better than 10-20 miles.
Let’s then take mobileye who doesn’t need any setup time other than maybe a few days (such as in Germany). Will their disengagement be better than 10-20 miles?
There again is your answer to whose ahead.
Educated Logic & Reason, not emotion and fandom.
Last edited:
Cheburashka
Active Member
Week 2 of 2021 and no significant improvements.
Nothing I see in the latest videos leads me to believe we are any closer to a general release than a month ago.
Silicon Desert
Well-Known Member
maybe yes, maybe no. I'm not thinking that they were all continuing to work through the holidays. Let's see what happens in the next 30 days.
powertoold
Active Member
Tesla is at least 5 years ahead IMO. Where's that BMW traffic light feature?
DanCar
Active Member
Does the current, I think it is release #9, still swerve with intermittent parked cars? That looked like the biggest issue to me. Makes you look like you've had too much Teslaquila
n.one.one
Info w/o Drama
The drunken swerving seems to be mostly resolved but the good thing is that it will still duck into open spaces to avoid oncoming cars when space is tight and the other vehicle has the right of way.Does the current, I think it is release #9, still swerve with intermittent parked cars? That looked like the biggest issue to me. Makes you look like you've had too much Teslaquila
The biggest problem issue seems to be unprotected left turns and pulling out into the road in a timid fashion which means it's stopping or creeping in a dangerous spot.
mikes_fsd
Banned
I have to put it here, mainly to admit - this was surprising to hear.
I thought that the rewrite was with all camera NN's stitched in, but it seems that only some are (probably all forward facing cameras)
https://twitter.com/elonmusk/status/1353667962213953536
I thought that the rewrite was with all camera NN's stitched in, but it seems that only some are (probably all forward facing cameras)
https://twitter.com/elonmusk/status/1353667962213953536
Last edited:
powertoold
Active Member
Welp, I guess this means the competition is deader than dead.
More snake oil. With Elon it's always, "when we do xxx then you'll see".
But you both vehemently disagreed and tried to shame @verygreen when he told you that this is the same networks that have been in since 2018 and some 2019 and nothing has changed. You refused to listen to facts and rather hype. Just like you are now you are still refusing facts and are hyping more.
These ppl are like fangirls with zero logical thinking. Its quite sad. How can ppl leave their life like that. its stunning. no ability to think for themselves. Ofcourse with Elon its more hype, hype hype. Wait till [this] , it will be AMAZING. BLOW YOUR MIND. QUANTUM LEAP. SlickySMOOTH. BLAH BLAH BLAH.
mikes_fsd
Banned
I do not know how, but I missed this Elon reply.....
src: https://twitter.com/elonmusk/status/1353564655088619520
powertoold
Active Member
I wasn't really involved in that conversation, but these aren't the same networks, as Elon said. They've been steadily moving their networks to the stitched version.
powertoold
Active Member
Well, we did see a quantum leap with the FSD beta.
Cheburashka
Active Member
Along with a quantum leap in the number of disengagements per mile
It seems pretty clear from FSD beta visualizations that BEV Net is taking in multiple individual camera network predictions to make a unified prediction, and some things wouldn't work well at all with individual camera predictions such as the intersection traffic flow.
I suppose if we take Karpathy's slide more literally, those 3 outputs of BEV Net are the only ones that have been "upgraded" to surround view. However, pretty sure that isn't the case as at least traffic light detection seems to use multiple cameras as well, but there are also plenty of other predictions from individual camera networks at top that might not have been upgraded yet, e.g., static objects (cones and road closures) and road signs (stop except right turn only)?
Here's some more things that the networks could predict and get upgraded to surround:
Dan Detweiler
Active Member
Similar threads
- Question
