Wiki MASTER THREAD: Actual FSD Beta downloads and experiences

[brunla]

So there is a reason I am not seeing many FSD Beta videos in LA. Honestly I was hoping 10.4 was going to be better but it's not. It is so dangerous in a city like LA. Within 2 minutes of turning it on I had to disengage. There was a light way ahead with a pedestrian on the corner. It phantom braked to a stop about 300 feet ahead of the crosswalk. If there was a car behind me they would have crashed into me. In LA people ride your tail. Between that and purposely trying to crash into other vehicles without warning, I do not see how this can work in a busy city with aggressive drivers like LA. When I look at videos of people smoothly driving, I do not understand. My experience is not like that at all. Maybe it is the new refresh S, or maybe LA, but I gotta turn it off and wait for a few updates before I try again. I understand it is "Beta", but autopilot is also "beta" and it's mostly trustable. This feels more alpha to me. Very saddened by this because I wanted to participate and help make it better.

 

[Phlier]

FYI, there are differences among speed bumps (narrrow) and speed humps (wider) and speed tables (even wider / raised pedestrian crossing):

Key:traffic_calming - OpenStreetMap Wiki

wiki.openstreetmap.org

People often just refer to all of them as a speed bump.

Thanks for that! I've never heard of a speed table!

 

[_jal_]

Nice catch!

Did you report it, by chance?

I did

 

[nvx1977]

So there is a reason I am not seeing many FSD Beta videos in LA. Honestly I was hoping 10.4 was going to be better but it's not. It is so dangerous in a city like LA. Within 2 minutes of turning it on I had to disengage. There was a light way ahead with a pedestrian on the corner. It phantom braked to a stop about 300 feet ahead of the crosswalk. If there was a car behind me they would have crashed into me. In LA people ride your tail. Between that and purposely trying to crash into other vehicles without warning, I do not see how this can work in a busy city with aggressive drivers like LA. When I look at videos of people smoothly driving, I do not understand. My experience is not like that at all. Maybe it is the new refresh S, or maybe LA, but I gotta turn it off and wait for a few updates before I try again. I understand it is "Beta", but autopilot is also "beta" and it's mostly trustable. This feels more alpha to me. Very saddened by this because I wanted to participate and help make it better.

yeah totally relate with the aggressive driving aspect. People criticized the safety score for penalizing them for other drivers' behavior. but this helped limit FSD in areas where driving conditions are pretty tough. I was in LA for 2 months this summer and got to experience a different flavor of aggressive driving than what I see here in MA/Boston area. Both of which aren't conducive for a fledgling FSD AI.

 

[EVNow]

FYI, there are differences among speed bumps (narrrow) and speed humps (wider) and speed tables (even wider / raised pedestrian crossing):

Key:traffic_calming - OpenStreetMap Wiki

wiki.openstreetmap.org

People often just refer to all of them as a speed bump.

Hmm ... they haven't included an important traffic calming feature - though unintentional - potholes.

 

[eli_]

A lot of times when I think a new FSD build has fixed something, it's really just always been down to a 50/50 coin flip. Like if it chooses the correct left or right lane when entering this one roundabout, the same build will choose the wrong lane half the time. So it's impossible to say if a build has improved anything until you have a large enough sampling and control for time of day, weather, traffic, and moon phases.

 

[aronth5]

Hmm ... they haven't included an important traffic calming feature - though unintentional - potholes.

Agree and those nasty steel road plates in construction zones where you need drive less then 10mph to survive.

 

[eli_]

One thing that impressed me today on 10.4, it moved over for a construction lane closure before there were even visible cones. The only indications were a literal "right lane closed" sign and a lane merge sign...

No traffic ahead so I don't think it was a speed-based lane change. And I know it's not route/nav-based because I drive this road multiple times per day and it doesn't ever change lanes.

 

[Mardak]

Agree and those nasty steel road plates in construction zones where you need drive less then 10mph to survive.

I actually got a construction steel plate to get visualized as a speed bump. But I didn't have time to slow down and take a picture with all the active construction going on. I had already dropped the speed to 15mph, and I don't think the car slowed down any more… (other than to very briefly want to drive straight into a temporary concrete barrier…)

 

[tivoboy]

Yes - may be a better option. I bought the mount first, and then realized need an extender.



I should start a click-bait thread with the title "How FSD Beta cost me $350".

Honestly, Tesla should have a program for high contributors that they send out cameras and a mount to so we can capture more of the what/where/why how that they can review.. the time stamps on a GoPro and the in car videos, and bug reports should lime it up easily. And, Dojo can process the entire Netflix library in 72 hours and categorize it all, so there’s that.

(okay, I just made up that last point. ;-0

 

[Mardak]

Last night/early morning ~20 snapshot usage seems to have resulted in 1.2GB of upload, which assuming ~400MB per clip was only 3 recordings sent back. This morning had ~10 snapshots uploading 1.6GB would then maybe be 4 recordings? Unless they changed what gets uploaded as part of FSD Beta 10.4 to be smarter in how much actually needs to be saved.

Anybody notice differences in FSD Beta 10.4 video snapshot uploads? Unclear if the extra callout in the release notes was because there was something changed or just an extra reminder.

In addition to the 10.4 Sunday early morning 1.2GB and Sunday morning 1.6GB, I've seen 6.8GB (17 recordings?) Sunday afternoon, 2.8GB (7 recordings?) Monday morning and 4.0GB (10 recordings?) Monday afternoon. I wonder if the early morning one dropped recordings or compressed really well because it was so dark outside.

Edit: Compared to the Y that got 10.4 Sunday night: 1.3GB (3 recordings?) Sunday afternoon, 1.2GB (3 recordings?) Sunday evening, 1.1GB (3 recordings?) Sunday night, 2.5GB (6 recordings?) Monday early morning -- pretty sure nobody drove, so the car decided to start uploading more? -- and 1.3GB (3 recordings?) Monday morning -- also pretty sure car hadn't been driven since last night.

 

[eli_]

Anybody notice differences in FSD Beta 10.4 video snapshot uploads? Unclear if the extra callout in the release notes was because there was something changed or just an extra reminder.

In addition to the 10.4 Sunday early morning 1.2GB and Sunday morning 1.6GB, I've seen 6.8GB (17 recordings?) Sunday afternoon, 2.8GB (7 recordings?) Monday morning and 4.0GB (10 recordings?) Monday afternoon. I wonder if the early morning one dropped recordings or compressed really well because it was so dark outside.

Edit: Compared to the Y that got 10.4 Sunday night: 1.3GB (3 recordings?) Sunday afternoon, 1.2GB (3 recordings?) Sunday evening, 1.1GB (3 recordings?) Sunday night, 2.5GB (6 recordings?) Monday early morning -- pretty sure nobody drove, so the car decided to start uploading more? -- and 1.3GB (3 recordings?) Monday morning -- also pretty sure car hadn't been driven since last night.

Not sure but I noticed a big drop in the amount of uploads around the 10.3 release. Often drives won't cause anything significant to get uploaded now unless I manually trigger a snapshot. From greentheonly's research, the triggers that cause a snapshot to be uploaded can be changed at any time without pushing a new build to the car. The car just periodically fetches the triggers from a JSON file on a server, so they could change multiple times per day even. Also they can configure a trigger to upload over cellular instead of waiting for Wifi.

 

[EVNow]

Anybody notice differences in FSD Beta 10.4 video snapshot uploads? Unclear if the extra callout in the release notes was because there was something changed or just an extra reminder.

In addition to the 10.4 Sunday early morning 1.2GB and Sunday morning 1.6GB, I've seen 6.8GB (17 recordings?) Sunday afternoon, 2.8GB (7 recordings?) Monday morning and 4.0GB (10 recordings?) Monday afternoon. I wonder if the early morning one dropped recordings or compressed really well because it was so dark outside.

Edit: Compared to the Y that got 10.4 Sunday night: 1.3GB (3 recordings?) Sunday afternoon, 1.2GB (3 recordings?) Sunday evening, 1.1GB (3 recordings?) Sunday night, 2.5GB (6 recordings?) Monday early morning -- pretty sure nobody drove, so the car decided to start uploading more? -- and 1.3GB (3 recordings?) Monday morning -- also pretty sure car hadn't been driven since last night.

How does this square with the notion only 5 recordings are kept for upload ... ?

 

[AlanSubie4Life]

10.4 seems about the same, as fully expected. Veering & ping-ponging abruptly between lanes on merges (hadn't done that before, new behavior on this trivial merge, but entirely possible it was just luck), slowing suddenly for flashing yellow overhead lights (for an old and now non-existent school zone), and phantom braking for shadows, as normal. Not enough stops to judge smoothness there (I expect the same - “butter smooth,” but think frozen butter here). Haven’t been able to have it successfully do my unprotected left yet (0 for 1 on attempts; disengaged because it did not advance crisply to the point of visibility, immediately, as a human would - traffic behind was present so I had to use a very strict disengagement policy). Cutting off corners in neighborhood as usual. Only 5 miles so far but a sufficient sample size to say it's about the same.

Overall exactly as expected. We probably have at least a year (at least!) to go before it shows some semblance of smooth driving style. Lots of tricky features to add, and lots of refinement. Maybe much longer, given the current state of NoA! (Especially since Elon has said they put a lot of work into NoA jerk minimization, yet it still can't minimize jerk!) Maybe there will be some sort of perception overhaul at some point which leads to much better results. Probably not though.

 

[nvx1977]

I noticed that my car stopped uploading anything toward the tail end of 10.3.1 if I didn't press the snapshot button. Seems like Tesla might be more aggressive with disengagement / intervention triggers the first week of their 2-week sprint.

 

[pitabun]

I'm getting quite used to stopping autopilot when I see other cars behind me.

If you already have your route set, doesn't FSD still gather data regardless of whether you use it?

 

[spokey]

As in you tapped the go pedal, and it didn't want to continue by itself? I've never experienced the car overriding my pedal presses to not budge, but I do keep it depressed if it feels like the car won't continue accelerating. I believe typically actual continued pedal presses are only overridden during automatic emergency braking and pedal misapplication mitigation -- both to avoid crashing or running off the road.

A note that if you do just keep pressing instead of disengaging, there is a chance that FSD Beta will suddenly decide on a different path.

I had that happen today. Just got 10.4. I've been letting it navigate a stupid way home which involves an unprotected left across a couple lanes of high speed traffic. Figured it was good training. A couple days ago it handled this just fine. Today it stopped. There was at least 1/4 mile clear (probably the light further up had changed) and it didn't go. Gave it a pedal and nothing. It still seemed to be in FSD (although I can't swear the icons were lit). I took it out of FSD, and put it back in as soon as I crossed the road. It was fine (well as fine as it gets) after that.

 

[spokey]

Anyone had their brake pads changed yet? Thought about it for 10.4 but looks like 10.4 is going to eat them up too so might as well wait.

Also the Beta is SOOOOOOO inefficient that Tesla may need to offer 2 EPA milage ratings in the future. One regular and one (WAAAAAAY lower) on FSD. Here is last weeks.

View attachment 730719

Strange I was going to post the opposite. Had around a 35 mile round trip this afternoon. I almost always set trip A when I leave the house. I noticed when almost home the Wh/mi was 234 and thought I could not beat that. That despite an increase in phantom braking.

 

[spokey]

Question about lane changing. I'd swear that the prior release allowed you to set the lane change to confirm. Is that not true? I was surprised that mine just changed lanes today without warning. When I checked the lane change options are grayed out while in FSD. At least when driving.

 

[AlanSubie4Life]

Strange I was going to post the opposite. Had around a 35 mile round trip this afternoon. I almost always set trip A when I leave the house. I noticed when almost home the Wh/mi was 234 and thought I could not beat that. That despite an increase in phantom braking.

For all the people making these comparisons, be sure to ensure you are comparing trips with the climate control OFF, before and after. It has to be entirely off (gray icon), otherwise the comparisons are essentially meaningless. Obviously has to be the same trip segment, similar ambient temperature, no rain or wet roads, etc., too. If you use the AC or heat a little bit, it will completely change the results.

Personally I would expect FSD to be slightly less efficient than an efficient driver, but by hovering over the brake pedal you can tell that actually it doesn't hit the brakes THAT much, and usually at low speed, where there is minimal energy lost. That being said, the sloshing of energy back and forth due to excessive regen isn't good and wastes lot of energy too, and that most certainly increases with FSD vs. an efficient driver.