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

FSD Beta Videos (and questions for FSD Beta drivers)

JulienW

Active Member
Jul 7, 2018
2,643
3,117
Atlanta
Couple new beta testers posting their first videos. Glad to see someone from Illinois now. Will be interesting when we actually get snow here (and the other midwest states).
My vote goes to @n.one.one to be included in the beta group as reward for keeping this thread up-to-date!
Yea would love to see an Atlanta Tesla represented. We have sooooo many on the roads. Kim from Like Tesla was one of the original Tesla Youtubers and might have got it but she decided to change her channel a couple of years ago.
 

turnem

Member
Apr 24, 2019
503
671
Charlotte, NC
Couple new beta testers posting their first videos. Glad to see someone from Illinois now. Will be interesting when we actually get snow here (and the other midwest states).
My vote goes to @n.one.one to be included in the beta group as reward for keeping this thread up-to-date!
Agreed! This thread is a treasure trove of information and it take a lot of time to keep up with all the videos!
 
  • Like
Reactions: n.one.one

aronth5

Long Time Follower
Supporting Member
May 8, 2010
2,806
1,764
Boston Suburb
I have a feeling a lot of the testers want to highlight edge cases where the FSD today struggles. Right now it is supposed to be autopilot of city streets, which means that it will let you relax while it does the boring driving, but whenever the situation get’s tricky it let’s the human drive. Just like autopilot does sometimes, when there are construction, bad weather etc. As more data is gathered the number of times a human needs to take over falls just like autopilot for highway has done. So imo it is more important to get a feel for how much the FSD actually makes the driving safer and more comfortable, not if there still are edge cases preventing the system from being 100% ready for LVL 4.

It is good for Tesla that users are finding edge cases, but it is less useful for us investors. So youtubers and forum members focusing on this seems misguided.

As for edge cases yeah, we should monitor them, but not obsess about them. The trend is important, not where we are at today.
Which videos and uploads do you believe Tesla's Development staff is most interested in? The straight word drives or the edge cases?
For me I'm much more interested in seeing edge cases when a new beta build comes out since that seems to be the best way to determine progress or lack of progress.
Regular drives have almost become boring with the continued improvement so keep those edge cases coming:)
 

n.one.one

Info w/o Drama
Oct 12, 2014
867
927
Metro Atlanta, GA, USA
Which videos and uploads do you believe Tesla's Development staff is most interested in? The straight word drives or the edge cases?
For me I'm much more interested in seeing edge cases when a new beta build comes out since that seems to be the best way to determine progress or lack of progress.
Regular drives have almost become boring with the continued improvement so keep those edge cases coming:)
I agree that the videos that focus on something specific (left turns, roundabouts, etc.), and edit properly seem to be the most useful at this stage. It's a shame that the today's construction focused video from Tesla Raj was on a previous release. I hope he does that again on a current release.
 

heltok

Active Member
Aug 12, 2014
1,302
11,616
Sweden
Which videos and uploads do you believe Tesla's Development staff is most interested in? The straight word drives or the edge cases?
For me I'm much more interested in seeing edge cases when a new beta build comes out since that seems to be the best way to determine progress or lack of progress.
Regular drives have almost become boring with the continued improvement so keep those edge cases coming:)
I doubt the Tesla engineers are spending hours watching youtube videos. I think they see enough videos uploaded from the vehicles from all cameras which is a lot more useful than a camera mounted inside the vehicle. The labellers will classify interventions and they get some statistical overview of what is the most common reasons for interventions with some severity added to them and then they make decision on where to focus their work and what data to request from the data engine.
 

aronth5

Long Time Follower
Supporting Member
May 8, 2010
2,806
1,764
Boston Suburb
I doubt the Tesla engineers are spending hours watching youtube videos. I think they see enough videos uploaded from the vehicles from all cameras which is a lot more useful than a camera mounted inside the vehicle. The labellers will classify interventions and they get some statistical overview of what is the most common reasons for interventions with some severity added to them and then they make decision on where to focus their work and what data to request from the data engine.
That wasn't the question I asked. Do you believe Tesla FSD developers are more interested in plain drives or edge cases?
 

Dan Detweiler

Active Member
Apr 21, 2016
3,079
13,313
Canton, Georgia
That wasn't the question I asked. Do you believe Tesla FSD developers are more interested in plain drives or edge cases?
Personally, I don't think they are watching these videos at all except to verify behavior patters they detect from their data. The cars are sending them all the pertinent data at all times in a format they can immediately use. They don't need to interpret a video to figure it out.

No source, no proof, just my gut feelings in the matter.

Dan
 

EVNow

Well-Known Member
Sep 5, 2009
10,454
30,495
Seattle, WA
That wasn't the question I asked. Do you believe Tesla FSD developers are more interested in plain drives or edge cases?
They will look for "edge cases" since plain drive examples are easy to get (they get it when they drive the cars themselves). Infact edge cases are oversampled when they train NN.
 
  • Like
Reactions: Matias

Dan D.

Member
Dec 7, 2020
775
967
Vancouver, BC
Hmm, 9 disagrees for my post urging caution

...
I would suggest they ban unprotected left, merging with high speed vehicles from a stop, turning through high speed traffic. None of these are being done safely yet.

and 6 likes for the post saying "There will always be accidents and deaths"

The problem is you're looking for perfection. There is no such thing. There will always be accidents and deaths, even with Level 5. The important thing is making it significantly safer than a human driver.

I use my non-beta autopilot and NoA every day on the highways because it makes me a safer driver. Let these people do what they have agreed to do. They're doing a damn good job

Dan

I know what you're saying that this is a Beta product, and as such it takes time, experimentation, accidents and feedback to improve the system. Hopefully you may also see my side which is nobody else on the road has agreed to this.

I'm really on your side, as if people had stopped Uber from their poor safety practices in Arizona there may not have been the first fatality, and Uber suffering business losses. Uber has now improved their safety based on all the NTSB recommendations. Why did it take that disaster to take the steps that were needed from the outset. Tesla is taking risks and I'm pointing them out.

edit: Uber, not Waymo
edit2: I'm curious why people press disagree. To which point?
 
Last edited:
  • Disagree
Reactions: WattPwr

heltok

Active Member
Aug 12, 2014
1,302
11,616
Sweden
That wasn't the question I asked. Do you believe Tesla FSD developers are more interested in plain drives or edge cases?
Yes, its good that they drive around in edge cases. But putting it on YouTube adds little to both Tesla and us viewers after it has been done a few times already...
 

Dan Detweiler

Active Member
Apr 21, 2016
3,079
13,313
Canton, Georgia
Hmm, 9 disagrees for my post urging caution



and 6 likes for the post saying "There will always be accidents and deaths"



I know what you're saying that this is a Beta product, and as such it takes time, experimentation, accidents and feedback to improve the system. Hopefully you may also see my side which is nobody else on the road has agreed to this.

I'm really on your side, as if people had stopped Uber from their poor safety practices in Arizona there may not have been the first fatality, and Uber suffering business losses. Uber has now improved their safety based on all the NTSB recommendations. Why did it take that disaster to take the steps that were needed from the outset. Tesla is taking risks and I'm pointing them out.

edit: Uber, not Waymo
edit2: I'm curious why people press disagree. To which point?
Do you feel like there WON'T always be accidents, no matter how advanced the system?

Dan
 

powertoold

Active Member
Oct 10, 2014
2,952
6,388
USA
Dan D. is one of those cancel culture types. One fatality with Autopilot, ban it.

We understand your concern, but your solution isn't practical.

And yes, Tesla is taking a (calculated) risk here. Time will tell if they've correctly assessed it.
 

Daniel in SD

Well-Known Member
Jan 25, 2018
7,043
10,458
San Diego
Do you feel like there WON'T always be accidents, no matter how advanced the system?

Dan
This is a silly straw man argument. Everyone agrees there will be accidents. Safety is relative. Would releasing beta FSD to everyone improve road safety or not?
I don't think it would though I have to admit I have a morbid curiosity. I can't really figure out what Tesla is up to. They could have easily found all these bugs in internal testing so it seems like there are two possibilities. This is mostly a marketing stunt or they're evaluating how well normal drivers can monitor the system (some can, some terrify me).
 
  • Like
Reactions: WattPwr and Dan D.

Tyler34

Member
Dec 11, 2019
414
3,069
Oregon
This is a silly straw man argument. Everyone agrees there will be accidents. Safety is relative. Would releasing beta FSD to everyone improve road safety or not?
I don't think it would though I have to admit I have a morbid curiosity. I can't really figure out what Tesla is up to. They could have easily found all these bugs in internal testing so it seems like there are two possibilities. This is mostly a marketing stunt or they're evaluating how well normal drivers can monitor the system (some can, some terrify me).
It is my opinion they've taken the guard rails off to train the NN during beta and will turn them back on for the wider release. For example there could've been a hard coded never go into an oncoming lane. The NN would never learn that's usually a no-no if it never got the opportunity to get corrected when it makes that decision. If you have a fence around your fireplace, your toddler will never learn to not touch the fire.
 

DanCar

Active Member
Oct 2, 2013
2,026
2,115
SF Bay Area
... nobody else on the road has agreed to this. ...
Nobody agrees to be hit by a drunk driver but it happens, perhaps every night. Nobody agrees to be a victim of road rage, but it happens.
Driving is dangerous. You can't remove the danger for the most part.
Driverless promises many advantages. To take the ultra paranoid approach that, oh my gosh somebody could die, even though it is a unlikely, is ridiculous. People are dying everyday. We should take steps to accelerate this technology, not slow it down as your posts suggest.
 
Last edited:

Products we're discussing on TMC...

About Us

Formed in 2006, Tesla Motors Club (TMC) was the first independent online Tesla community. Today it remains the largest and most dynamic community of Tesla enthusiasts. Learn more.

Do you value your experience at TMC? Consider becoming a Supporting Member of Tesla Motors Club. As a thank you for your contribution, you'll get nearly no ads in the Community and Groups sections. Additional perks are available depending on the level of contribution. Please visit the Account Upgrades page for more details.


SUPPORT TMC
Top