2017.50.3.f3425a1 is out!

[Daniellane]

Maybe we are reading two different things? As I said earlier, it adjusts at what point the wipers are triggered. Level 1 waits a little longer (more rain buildup on the windshield) and level 2 is quicker on the trigger.

Ok,
I think it is stated clearly, although it’s wordy:

“Want to turn them on intermittently when a medium amount of rain has accumulated? Choose Auto 1 option on the wiper lever”
Or choose the more sensitive Auto 2 option if you prefer wiping as soon as the vehicle senses just a light mist or drizzle, such as when you’re driving out of the garage on a foggy morning. To make sure the windshield is clear, the wiping speed will automatically adjust based on how much rain accumulates between each wipe.

 

[Daniellane]

50.3 will do this too. Earlier versions before .42 also did this. It will detect the car, start slowing (regen), then accelerate (lose sight?) and then hard stop (physical brakes) (if you let it keep going). I usually tap the brakes at the lose sight stage and let regen slow me (if available, recent cold snap has also reduced my ability to test).

Darn, thought it was fixed.
Right you are. Tested again today on 50.3 and “Russian roulette suicide mode” still exists in some instances of a distant stopped vehicles. Had to disengage as it got too close for comfort.

 

[BigD0g]

Russian roulette suicide mode is really the best way to phrase it and oh so true.

 

[NewTMSMan]

Your test seemed perfect. I was testing 17.50.2 over the weekend and it appeared to work better going to the right (like your test) then going to the left. Any chance you tested going both left and right auto lane changes?

I tested 2017.50.3 on the highway today with multiple lane changes left and right and it is much much better than 2017.50.2. Lane changes were smooth, but not too slow. They just about have it right. Also the car maintained a good lane position at almost all times, in prior versions it seemed to close to adjacent lanes when passing.

 

[BlueRocket]

I just measured at 160-180 kb/s so still slow.

I believe that someone reported that AT&T was doing the throttling and Tesla was aware of it. Unfortunately, firmware cannot fix the issue if AT&T is slowing LTE.

 

[d21mike]

Here’s the video of multiple 50.3
Auto Lane Changes.
Right - left - right...
Quick, Precise & Smooth in both directions

Good Test. Thanks for the feedback. I went out and did some testing myself but without the video. I did notice an improvement but I think your test seems a bit smoother. I also noticed that if I have my TACC set at a higher speed then the traffic I noticed it will speed up while changing lanes instead after it is complete. I actually like that. I have noticed it in the past so not sure it is much different.

 

[SucreTease]

I concur with those saying AP is better. Lane changing is super smooth now and overall it feels a bit more confident than 50.2. I haven't done any longer trips yet, but I'm hopeful it'll be an improvement.

I just got 50.3 this morning. In stop-and-go traffic I noticed that it seemed to be doing a much smoother job braking instead of the harder braking it used to do. Also seemed to be better at speeding up when the car in front starts to go.

 

[croman]

I just got 50.3 this morning. In stop-and-go traffic I noticed that it seemed to be doing a much smoother job braking instead of the harder braking it used to do. Also seemed to be better at speeding up when the car in front starts to go.

Still variable. I've driven over 400 miles so far on it and it gets it more right but its still oddly erratic.

Also lane changes can be done in much tighter situations and it tracks cars and they appear as cars if they even adjacent to you (the car presumably saw it further out but still tracks it and displays it during a lane change).

It also handles cars entering your lane better (recognizes them sooner and is a lot less jumpy when it happens (uses regen to smoothly decelerate).

 

[Swift]

Thanks for pointing out the obvious

You’re welcome.

 

[Daniellane]

Interesting. Looks like this is going to AP2 cars almost exclusively. Which isn’t surprising. May not have any changes in code that effects pre-AP2 cars. Only went to 4 non-autopilot/AP1 cars out of 332 total.
I’m guessing those were pushed by service centers.

 

[chillaban]

So now that I put a couple dozen miles on .50.3, some general thoughts:

Autowipers:
It was only drizzling slightly in my dark commute home. It wiped a few times. Definitely not as eager as when it's in broad daylight but it does seem to somewhat work. Overall no complaints so far. It's not perfect yet but it's a great start and alleviates 90% of the use cases of automatic wipers.

Autopilot:
There is a definite improvement in .50.3. I did not have enough time in .50.2 to make an assessment other than seeing it follow a cloverleaf onramp that it was never able to do before. But in .50.3, I saw a noticeable jump in performance compared to .48, and it fits very nicely in the story of .34 to .42 to .48 -- really feels like the next chapter. More concretely:

- It navigated long city intersections without lane lines reliably, not requiring a single disengagement
- It seems to respect lane lines better on hilly curves. Once it did get uncomfortably close to an edge but it does feel improved compared to before.
- Lane lines appeared much more stable compared to before. I swear there's a neural net change since .48.
- Stationary cars are recognized from afar and move more "smoothly" across the screen
- When moving between lanes manually in the city, the lane lines indicated on the IC seem much more "correct" than before in terms of where the car is relative to a lane line. Before, in the city, it sometimes showed completely nonsense lines or go from gray to black when straddling a line.


I still need more time, but this release definitely feels like it contains AP2 improvements.

 

[ecarfan]

I got this firmware update for my 2017 Model X and had the opportunity today to test the auto wipers in light rain. They appeared to work as well as the auto wipers on my 2013 non-AP Model S that I owned for 3 1/2 years. Seem fine to me. Not perfect on either car, but very usable.

 

[Bebop]

It's interesting that this is still a week 50 release. I guess this might be subtle validation that the different software teams are operating independently.

Yeah kind of weird it still says 2017 on the release name as well.

 

[PrGrPa]

Doesn't seem to. What seems to change is the interval of wiping. The best I got were two consequtive wipes, mostly it was one wipe and then it sort of stopped for a few seconds.

Wait and look... wait and look... wait and look... oh, rain! WIPE!... wait and look... wait and look... oh, rain! WIPE! WIPE!... wait and look... wait and look... wait and look... wait and look... wait and look... oh, rain! WIPE!... wait and look...

Sounds like it’s emulating the way a human would wipe the screen when presented with rain that wasn’t enough for the slow or fast wiper settings. I like the sound of this.

 

[AnxietyRanger]

Sounds like it’s emulating the way a human would wipe the screen when presented with rain that wasn’t enough for the slow or fast wiper settings. I like the sound of this.

Except the rain was steady moderate rain. Not light rain, steady moderate rain.

 

[AnxietyRanger]

I got this firmware update for my 2017 Model X and had the opportunity today to test the auto wipers in light rain. They appeared to work as well as the auto wipers on my 2013 non-AP Model S that I owned for 3 1/2 years. Seem fine to me. Not perfect on either car, but very usable.

Daylight or darkness?

My pre-AP Model S rain sensor worked perfectly fine, I have no complaints. This one, though, nothing like it. Erratic would be how I'd describe it during darkness. Hopefully I will get some daytime rain driving time soon to test that.

 

[AnxietyRanger]

Definitely not as eager as when it's in broad daylight but it does seem to somewhat work.

I don't have daytime experience yet of rain, other than the episode of ghost wipe coming out of a tunnel, but this sounds accurate regarding darkness... It somewhat works.

 

[banned-66611]

Anyone else with a similar experience? Auto-wipers look completely broken, and totally unsafe in this video.

 

[saniflash]

I have the feeling that automated high beams is working much better on my AP2 car now than right after the initial release. It's actually very reliable now, I only have to manually disengage maybe one time over a 40 miles stretch Autobahn. Also, if you look at forum posts, there's hardly anyone complaining anymore.

I'm optimistic that we'll see a similar improvement with the rain sensor over the next releases. Also, is there any reason to assume that no calibration has to take place initally?

 

[AnxietyRanger]

I'm optimistic that we'll see a similar improvement with the rain sensor over the next releases.

Of course we will see improvements.

The question is the same as with AP2 in general: how soon and how well, not if. But the how soon, how well question has historically been offering a wider variance of answers than perhaps is comfortable to many.