FSD Beta 10.11

[jebinc]

just curious... he indicates that his videos are unedited and raw cut. Most of his video are actually as he states as silky smooth and perfect. I live in the bay area and may post exactly the same routes that he is using to see if the results are the same.

Who knows. For all we know, Tesla optimizes for routes routinely taken by Elon and shills. Cheap PR and a happy uninformed leader. Who knows.

 

[Brendazzle]

it happens every single time I get into the car if my phone isn't set to silent.
As soon as the phone connects there is a 100% volume "ding" from the car as the phone connects to the car.
If the phone is set to silent there is no ding.
If the phone isn't on silent the connect noise is always 100% volume regardless of car or phone volume setting.

Elon says he has never heard of the issue - probably because there is no worthwhile support channel to report issues any longer - unless you open a service call to report every software bug, or get an "influencer" to tweet at Elon.

I had this happen once, switched from radio to Bluetooth and it came on full volume, I was not happy. It occurred to me that with the Bluetooth connection the phone controls the volume, which it seems to use as a default starting point rather then the cars volume setting. I’m not sure that’s exactly what it is, but it needs to be fixed.

 

[EVNow]

Who knows. For all we know, Tesla optimizes for routes routinely taken by Elon and shills. Cheap PR and a happy uninformed leader. Who knows.

One thing that definitely happens is the maps are correct in SF and areas where Elon drives. They may also be able to make sure the planner handles some of those common routes well - not just for Elon but for the AP team and the internal testers (i.e. not employee, but full time AP team internal testers). All this can happen without them even realizing it ....

As I have said before - if roundabouts and school zones are handled properly - a large % of my drives will be intervention free too.

I think people who use FSD Beta rarely have more interventions because they are more likely to intervene at the slightest hint of the car not driving like the way they do themselves. As people use FSD Beta more - they have less interventions as they get used to how FSD Beta drives. There is a learning curve here.

This is self-reinforcing. People who don't get over this learning curve, will continue to have lot of intervensions and will not use FSD Beta much as a result. So, they never get used to FSD Beta and will continue to believe it to be very bad. On the flip side, those who use it often learn how FSD Beta drives and intervene less as a result. And they start feeling FSD Beta drives fairly well - with "known issues". I think this is where I am. So, you almost see those two "camps" here - those who use FSD Beta all the time and those who rarely use it.

But, every time I go on a drive I've used FSD Beta before, I tend to have more interventions, as I'm unsure what its going to do. After a few drives on that route the interventions are lesser as I know how it handles those routes.

 

[Reddy Kilowatt]

Did some more driving around in the rain and FSD continued to work. Still no Psycho wipers, still smooth and accurate.

Confidence level is up with this version, I'm letting the leach get a little looser.

 

[cobra]

I had this happen once, switched from radio to Bluetooth and it came on full volume, I was not happy. It occurred to me that with the Bluetooth connection the phone controls the volume, which it seems to use as a default starting point rather then the cars volume setting. I’m not sure that’s exactly what it is, but it needs to be fixed.

iPhone? I noticed some volume weirdness after the most recent updates. Car has not changed, so probably Apple again here.

 

[wknickless]

As I have said before - if roundabouts and school zones are handled properly - a large % of my drives will be intervention free too.

+1 on roundabouts needing work. Maybe we can lobby Northern California to put in more roundabouts?

 

[PACEMD]

Maybe we can lobby Northern California to put in more roundabouts?

That's a great idea. Could make it like Spain. Spain is one big 6 lane roundabout with a little straightaway thrown in just for kicks.........

 

[aronth5]

Did some more driving around in the rain and FSD continued to work. Still no Psycho wipers, still smooth and accurate.

Confidence level is up with this version, I'm letting the leach get a little looser.

Do you use it all the time or do you restart using FSD when you get a new release ?

I don't particularly remember first few drives as being bad and it improving over time. Probably because when I get a new release I go through my most traveled routes to see the differences.

We have talked about this before - I think there is a lot of variability even with the same release - because of the way NN works and ofcourse traffic & weather conditions. So, to really figure out what is happening we have to observe over a period of time. Best would be to note down all the interventions/disengagements and any other important observations (time it takes for the car to negotiate an empty intersection or roundabout etc). I keep thinking of documenting this - but I've never gotten around to doing it. May be from next release.

I wish Tesla would let us see our individual stats (even if they don't want to publish aggregates).

I use FSD for most drives especially long drives. There are several intersections/roundabouts where FSD is awful so I either disengage or use the accelerator to push thru. I try several standard routes that I will try out as soon as I get the latest build. I always take these routes 2-3 times since they usually get better and I always wait 2 weeks before making a judgement about the release. Hoping this won't be a regression.

 

[Sigma4Life]

I just drove to work making two stops along the way. home-Starbucks-Post office-work.

I ran from driveway to driveway on these three stops, and let it do its thing to point of failure and ignored other traffic tailgating me. I used the flashers through the traffic circle. I was determined not to touch anything on this trip. no turn signals, gas pedal bumps etc.

Car did great, but missed one lane change and tried to turn left from the right hand lane and I had to take over. Handled all the stop signs, circles, intersections exit ramps fine for over 10 miles. took me from driveway to driveway each time.

Stops signs are getting better, I would say it the car goes one car length past the stop bar then goes for it. 11.1 did not do that it would stall at that point and I needed to give it gas. I also got through my right on red to 55 MPH OK. same thing about a car length and the car commits and accelerates now, unlike 11.1.

Exit ramps continue to smooth out and eliminate the jerky braking and acceleration. worked my way around slow truck traffic also. also accelerated to merge into highway traffic. I let the car drive its way to the locations which was the correct way to travel. Turns are also smoothing out, no go halfway and aim for the centerline. It knows the turns now lines or not.

The problem is you’re giving it gas.

 

[LowlyOilBurner]

I bet I would have a lot more intervention-less drives if the roads by me we in better shape. FSD likes to blast into potholes, and I’m not into replacing cracked wheels and tires that have rips in the sidewalls.

 

[sleepydoc]

Do you use it all the time or do you restart using FSD when you get a new release ?

I don't particularly remember first few drives as being bad and it improving over time. Probably because when I get a new release I go through my most traveled routes to see the differences.

We have talked about this before - I think there is a lot of variability even with the same release - because of the way NN works and ofcourse traffic & weather conditions. So, to really figure out what is happening we have to observe over a period of time. Best would be to note down all the interventions/disengagements and any other important observations (time it takes for the car to negotiate an empty intersection or roundabout etc). I keep thinking of documenting this - but I've never gotten around to doing it. May be from next release.

I wish Tesla would let us see our individual stats (even if they don't want to publish aggregates).

I’ve posted this before, but shortly after my last FSD update I also got a map data update. The map data update happened automatically without my input, so had I not happened to look at the software page I wouldn’t have known about it. I definitely saw an improvement after the map update, though. It’s quite possible this happens to others without their knowledge.

Who knows. For all we know, Tesla optimizes for routes routinely taken by Elon and shills. Cheap PR and a happy uninformed leader. Who knows.

I think it’s a pretty safe assumption that areas frequented by Elon and the software developers get more attention than the rest of the U.S.

 

[sleepydoc]

One thing that definitely happens is the maps are correct in SF and areas where Elon drives. They may also be able to make sure the planner handles some of those common routes well - not just for Elon but for the AP team and the internal testers (i.e. not employee, but full time AP team internal testers). All this can happen without them even realizing it ....

As I have said before - if roundabouts and school zones are handled properly - a large % of my drives will be intervention free too.

I think people who use FSD Beta rarely have more interventions because they are more likely to intervene at the slightest hint of the car not driving like the way they do themselves. As people use FSD Beta more - they have less interventions as they get used to how FSD Beta drives. There is a learning curve here.

This is self-reinforcing. People who don't get over this learning curve, will continue to have lot of intervensions and will not use FSD Beta much as a result. So, they never get used to FSD Beta and will continue to believe it to be very bad. On the flip side, those who use it often learn how FSD Beta drives and intervene less as a result. And they start feeling FSD Beta drives fairly well - with "known issues". I think this is where I am. So, you almost see those two "camps" here - those who use FSD Beta all the time and those who rarely use it.

But, every time I go on a drive I've used FSD Beta before, I tend to have more interventions, as I'm unsure what its going to do. After a few drives on that route the interventions are lesser as I know how it handles those routes.

I think there’s a bit of a bimodal distribution. On roads where I’ve used FSD a lot I know exactly where it will mess up and will prophylactically disengage to avoid the hassle and to avoid annoying and inconveniencing the drivers around me.

 

[Twiglett]

just curious... he indicates that his videos are unedited and raw cut. Most of his video are actually as he states as silky smooth and perfect. I live in the bay area and may post exactly the same routes that he is using to see if the results are the same.

He is technically correct and I could post similar results, but that would be 1% of my driving. The other 99% would be disengage, takeover, stupid behavior drives
I'm pretty sure nobody would maintain "influencer" status if they did that though.

 

[aronth5]

I bet I would have a lot more intervention-less drives if the roads by me we in better shape. FSD likes to blast into potholes, and I’m not into replacing cracked wheels and tires that have rips in the sidewalls.

Agree. This time of year the percentage of disengagements for potholes is easily the largest percentage of my disengagements. Lets hope Tesla resolves this by early next year when the frozen roads melt and re-freeze creating a mess.

 

[sleepydoc]

Agree. This time of year the percentage of disengagements for potholes is easily the largest percentage of my disengagements. Lets hope Tesla resolves this by early next year when the frozen roads melt and re-freeze creating a mess.

As spring starts to come to MN I’ve been thinking about this, too. Level 6 autonomy is the ability to find the serpentine route through the pothole labyrinth that does the least amount of damage to your suspension.

 

[EVNow]

He is technically correct and I could post similar results, but that would be 1% of my driving. The other 99% would be disengage, takeover, stupid behavior drives
I'm pretty sure nobody would maintain "influencer" status if they did that though.

I think he has figured out the routes to drive to get good drives. He just drives them .... and given how many drives he posts - he definitely has a much better % of good drives than your 1%

BTW, he has also started showing the accelerator in the videos (he moved the camera back, I guess) - so we know if he intervenes as well.

One thing for sure is - he doesn't care about driving etiquette - he will not intervene if it is just inconveniencing the fellow drivers like we do.

 

[Supcom]

He is technically correct and I could post similar results, but that would be 1% of my driving. The other 99% would be disengage, takeover, stupid behavior drives
I'm pretty sure nobody would maintain "influencer" status if they did that though.

Watch Chuck Cook's videos, especially his unprotected left turn and unmarked road series. Despite showing how poorly FSD beta handles these situations, he still seems to get early releases.

 

[HappyPanda]

Looks like there may be hope yet for 10.10.2 users getting the latest update! (Sadly no reports of any updates from non-beta still)

https://twitter.com/x/status/1511465849235726339

 

[black knight 3]

One thing that definitely happens is the maps are correct in SF and areas where Elon drives. They may also be able to make sure the planner handles some of those common routes well - not just for Elon but for the AP team and the internal testers (i.e. not employee, but full time AP team internal testers). All this can happen without them even realizing it ....

As I have said before - if roundabouts and school zones are handled properly - a large % of my drives will be intervention free too.

I think people who use FSD Beta rarely have more interventions because they are more likely to intervene at the slightest hint of the car not driving like the way they do themselves. As people use FSD Beta more - they have less interventions as they get used to how FSD Beta drives. There is a learning curve here.

This is self-reinforcing. People who don't get over this learning curve, will continue to have lot of intervensions and will not use FSD Beta much as a result. So, they never get used to FSD Beta and will continue to believe it to be very bad. On the flip side, those who use it often learn how FSD Beta drives and intervene less as a result. And they start feeling FSD Beta drives fairly well - with "known issues". I think this is where I am. So, you almost see those two "camps" here - those who use FSD Beta all the time and those who rarely use it.

But, every time I go on a drive I've used FSD Beta before, I tend to have more interventions, as I'm unsure what its going to do. After a few drives on that route the interventions are lesser as I know how it handles those routes.

The maps may be correct in SF where Andrej Karpathy and other developers drive their Tesla's but as far as I know Elon packed his bags and left California some time ago. One would think Austin, Texas will get "enhanced" attention from Elon and route planning..

 

[EVNow]

The maps may be correct in SF where Andrej Karpathy and other developers drive their Tesla's but as far as I know Elon packed his bags and left California some time ago. One would think Austin, Texas will get "enhanced" attention from Elon and route planning..

Elon is back to "couch surfing" in Bay area

..said that not having his own place in the Bay Area from 2002 to 2017 forced him to sleep at Tesla’s Fremont, Calif. factory or in a hotel or friend’s spare bedroom.

Elon Musk describes ‘bleak’ life in ‘strange Gatsby-like house’

Elon Musk over the weekend reminisced fondly on his days couch surfing at the homes of Google founders Larry Page and Sergey Brin, and bemoaned the “strange Gatsby-like house” he is sta…

nypost.com