Firmware 8.0

[jgs]

Some of that takes time to surface. The NTSB and NHTSA investigations into the Joshua Brown case is still ongoing, and it's unclear what if any requests were made. Eventually they will become part of a publicly released report, but it's still possible this is a regulatory matter that's been freshly discussed, or it's possible that this is just Tesla acting out of fear of legislation/liability, or genuinely trying to experiment with different AutoPilot behaviors to see their impact on safety.

Yep, I tried to capture that in "fear of legislation" (should have added "or regulation" since plenty of rules can be made under the aegis of existing laws). But unless I'm very much mistaken, Tesla couldn't have been outright compelled to make changes without a rulemaking, which involves publication, comment period, etc.

 

[drklain]

Well Tesla has not told us why they made changes, so no I can't quote a law or regulation they have changed because. I can point you to https://www.transportation.gov/sites/dot.gov/files/docs/AV policy guidance PDF.pdf as well as other discussions in these forums about the fact the AP1 cars are now apparently slowing to 50mph when they detect a stopped vehicle in an adjacent lane (which happens to comply with state laws about passing a stopped vehicle...someone in one of the threads quoted the appropriate laws in several states).

As you say, until Tesla tells us why they made a change, we are speculating. I have certainly tried to make clear when I was speculating and when I was providing facts.

 

[pchilds]

Tesla needs to add a button to report an incorrect speed limit in google maps.

 

[WannabeOwner]

Tesla needs to add a button to report an incorrect speed limit in google maps.

Is there any other way to report that (e.g. via Google?) I wonder if correcting it, via Waze, comes to the same thing? I've done the later for my local roads, as its very rural here and I may be the only Waze user on some of them

 

[drklain]

Yes -- you can report errors directly to Google. Here's a link to the Google answer on how to do it:

Steps to report an error on the map - Computer - Google Maps Help

 

[jgs]

Is there any reason to believe Tesla is sourcing their speed limit data from Google?

 

[drklain]

As I understand it, the speed limit displayed comes from one of two places:

(1) the internal GPS database
(2) what the cameras read off of speed limit signs.

I am not sure if the GPS database it uses is the Google one for the main screen or the Navtek one that drives the turn-by-turn displays on the dashboard display above the steering wheel. Both of those systems have speed limit data in them but I don't know which Tesla is using.

In our minivan (which has dashboard nav), it displays current speed limit which is clearly coming from the onboard nav database (on an SD card) since the car has no cellular/internet connectivity driving nav.

I

 

[Alex D]

Well Tesla has not told us why they made changes, so no I can't quote a law or regulation they have changed because. I can point you to https://www.transportation.gov/sites/dot.gov/files/docs/AV policy guidance PDF.pdf as well as other discussions in these forums about the fact the AP1 cars are now apparently slowing to 50mph when they detect a stopped vehicle in an adjacent lane (which happens to comply with state laws about passing a stopped vehicle...someone in one of the threads quoted the appropriate laws in several states).

As you say, until Tesla tells us why they made a change, we are speculating. I have certainly tried to make clear when I was speculating and when I was providing facts.

You mean this?

 

[chillaban]

So... attempting to go back on topic: I'm not sure if it's just me, but I'm actually noticing substantial Autosteer stability and TACC improvements in 2.50.114 vs my previous 2.42.40 build. I didn't get any of the intermediate ones so I can't say for sure this is the build that introduced it.

Namely, a lot of the 8.0 wobbliness to Autosteer and tendency to depart the lane around highway curves seems gone. Now Autosteer is reminding me a lot of how it used to be in 7.1.

I thought TACC regressed in 2.42.40 compared to the original 8.0 build in suddenly stopping traffic. I had 2 cases under .42.40 where a sudden freeway slowdown resulted in a FCW alarm and I had to take over braking. I tried 2.50.114 in some of my typical heavy freeway traffic and it was a lot better about braking more aggressively when approaching slower traffic.

Just for kicks, I carefully supervised AP going on El Camino Real (technically a highway but quite frankly it's a dense city road), which is my benchmark torture test for AP. It handled traffic quite well. Prior to 8.0, attempting to use AP on this road was not for the faint of heart.


I'm pretty sure this isn't placebo effect but of course I don't have any scientific metrics for this.

 

[msnow]

So... attempting to go back on topic: I'm not sure if it's just me, but I'm actually noticing substantial Autosteer stability and TACC improvements in 2.50.114 vs my previous 2.42.40 build. I didn't get any of the intermediate ones so I can't say for sure this is the build that introduced it.

Namely, a lot of the 8.0 wobbliness to Autosteer and tendency to depart the lane around highway curves seems gone. Now Autosteer is reminding me a lot of how it used to be in 7.1.

I thought TACC regressed in 2.42.40 compared to the original 8.0 build in suddenly stopping traffic. I had 2 cases under .42.40 where a sudden freeway slowdown resulted in a FCW alarm and I had to take over braking. I tried 2.50.114 in some of my typical heavy freeway traffic and it was a lot better about braking more aggressively when approaching slower traffic.

Just for kicks, I carefully supervised AP going on El Camino Real (technically a highway but quite frankly it's a dense city road), which is my benchmark torture test for AP. It handled traffic quite well. Prior to 8.0, attempting to use AP on this road was not for the faint of heart.


I'm pretty sure this isn't placebo effect but of course I don't have any scientific metrics for this.

Yes I noticed an improvement too.

 

[jgs]

I'm actually noticing substantial Autosteer stability and TACC improvements in 2.50.114

Agreed, and also better CID stability -- routes that used to semi-reliably cause a reboot (no, really) now don't.

 

[Cyclone]

Is there any other way to report that (e.g. via Google?) I wonder if correcting it, via Waze, comes to the same thing? I've done the later for my local roads, as its very rural here and I may be the only Waze user on some of them

I don't know about reporting speed limits, but when visiting Iceland in 2016, I used the option of Waze to "pave" an unknown road. Did a whole 35 miles of it before entering Keflavik that Waze knew nothing about!

 

[TrafficEng]

So... attempting to go back on topic: I'm not sure if it's just me, but I'm actually noticing substantial Autosteer stability and TACC improvements in 2.50.114 vs my previous 2.42.40 build. I didn't get any of the intermediate ones so I can't say for sure this is the build that introduced it.

Namely, a lot of the 8.0 wobbliness to Autosteer and tendency to depart the lane around highway curves seems gone. Now Autosteer is reminding me a lot of how it used to be in 7.1.

I thought TACC regressed in 2.42.40 compared to the original 8.0 build in suddenly stopping traffic. I had 2 cases under .42.40 where a sudden freeway slowdown resulted in a FCW alarm and I had to take over braking. I tried 2.50.114 in some of my typical heavy freeway traffic and it was a lot better about braking more aggressively when approaching slower traffic.

Just for kicks, I carefully supervised AP going on El Camino Real (technically a highway but quite frankly it's a dense city road), which is my benchmark torture test for AP. It handled traffic quite well. Prior to 8.0, attempting to use AP on this road was not for the faint of heart.


I'm pretty sure this isn't placebo effect but of course I don't have any scientific metrics for this.

Yep, the AP1 performance does seem much improved since initial 8.0. It is tough to tell whether it is from the software update, or the HD maps have accumulated (in a mysterious way!) data which improves the experience. One even gets used to the speed limitation; the 50mph limit when passing stopped vehicles in adjacent lanes; and the arbitrary way AP1 decided whether to limit to speed limit or not (about 50/50 on arterial roads in my area). However, there are two general situations where you had better be extra ready to take over:
1) any 'blind rise' situation, but especially where the vertical curve is associated with a horizontal curve. All bets are off as to what AP1 will do.
2) approaching stopped vehicles ahead, around a curve. AP1 will generally (but not always) 'see' these vehicles very late, and do violent late breaking. Knowing this, I always intervene before we test the ultimate limits.

But all in all, 10,000 miles of great experience! Being surrounded by sensors, with the technology to help you drive, makes a much safer driving experience. Bring on 8.1!

 

[TrafficEng]

It's not a valid handle on the Teslamotorsclub.com forum....Kind of surprising someone didn't try to register with it!

How about this suggestion? We all just use @tesla as a handle for directing comments to Tesla. Sounds like a reasonable 'crowd sourced' way to direct owner input to them. You listening, @tesla?

 

[CHG-ON]

With that thought process all autonomous features should be ninnied that could impede our safety. Auto high beams, auto wipers, auto emergency braking, REGULAR cruise control.....where do we begin and end with the nannying...

I know. But as an old boss of mine said back in the eighties. Never underestimate the importance of the American. Sucks, really.

 

[LastGas]

The navigation maps Tesla uses are from Garmin. That's where a speed limit would come from. There is a Garmin error reporting page here: Garmin | Report a Map Error and it may be possible that you can interactively correct a map that eventually goes to Garmin here: HERE - Map Creator

 

[MP3Mike]

Of course it appears that Tesla only updates the maps once a year and they are a year old. So at a minimum it would take two years for a correction to get to your car. Hopefully that will change see since point.

 

[shonline]

Of course it appears that Tesla only updates the maps once a year and they are a year old. So at a minimum it would take two years for a correction to get to your car. Hopefully that will change see since point.

I've had two map update notifications since last March when I got my Model S. FYI.

 

[msnow]

I've had two map update notifications since last March when I got my Model S. FYI.

The March one you got was the Q4 2015 and the one you just got is the Q4 2016. That's once per year like most of us. Each of those took a year to develop so as @MP3Mike says the newest are already a year old.

 

[shonline]

The March one you got was the Q4 2015 and the one you just got is the Q4 2016. That's once per year like most of us. Each of those took a year to develop so as @MP3Mike says the newest are already a year old.

Gotcha - thanks. Why so late for my Q4 2016 vs this one? Just earlier this year?