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

Software Update 2018.39 4a3910f (plus other v9.0 early access builds)

This site may earn commission on affiliate links.
Why can't the drive-on-nav confirmation use a slight turn of the wheel to that direction to confirm instead of having to turn on the lane signal? That would be so much easier and less "naggy".
because people are stupid and deliberately make videos of themselves doing stupid things with autopilot.
That's why we can't have nice things - because some youtube %*^*# value views over everything else
 
Shouldn’t blind spot detection tell you if the next lane is occupied before you turn on the blinker? If it works as described, you first cause an emergency brake chain reaction on the next lane and then know better...

Any theories why dashcam is limited to ap 2.5?

Add all subversions and localisations, the forking in Tesla OS is insane.
Yes, it's a forking nightmare :)
 
Lane Change is not what makes a system L3 and Lane Change is not DDT. Its a feature. The SAE calls it "DRIVING AUTOMATION SYSTEM FEATURE"

I never said that V9 was L3 because it could do lane changing. I said V9's ability to monitor the environment and to know WHEN to do a lane change brings it one step closer to L3.

Lane change is not however DDT.

Ok, a lane change is not a DDT itself but it is requires a DDT. In order words, you can't do a lane change without performing a DDT, right? But again, I never said that the act of lane changing makes V9 closer to L3. I said that making a decision whether or not to lane change brings V9 closer to L3. V9 knows WHEN to make a lane change which pre-V9 could not do. Very different!

DYNAMIC DRIVING TASK (DDT) is All of the real-time operational and tactical functions required to operate a vehicle in on-road traffic, excluding the strategic functions such as trip scheduling and selection of destinations and waypoints, and including without limitation:

1. Lateral vehicle motion control via steering (operational);
2. Longitudinal vehicle motion control via acceleration and deceleration (operational);
3. Monitoring the driving environment via object and event detection, recognition, classification, and response preparation (operational and tactical);
4. Object and event response execution (operational and tactical); Maneuver planning (tactical); and Enhancing conspicuity via lighting, signaling and gesturing, etc. (tactical).

And does V9 not perform these DDT better than before?

What makes a system L3 is that the system is monitoring the environment (which is apart of the DDT), and the ability to warn the driver in advance because of that, while giving him sufficient time (for example 5-10 seconds) to intervene.

And does V9 not bring autopilot closer to this?
 
  • Like
Reactions: rush6410
Same here! It was 36.2:mad:

Yup. Last night I got a firmware update prompt, and the forum was saying v9 was in general release, so I went ahead and installed, though I've been holding out on 2018.14.1 for a few months because nothing since offered me anything worthwhile to balance the greater nags. I'm now the proud possesser of 34.1. Oh, well.
 
How do you know when it pushes to mass people? When more than a few get it. I dont understand how to read the firmware tracker. So only 3 cars got the update?

it works like this:

People voluntarily register their VINs and cars etc to a site like TeslaFi, and you give them access to read your firmware version from the API, or your report it yourself manually.

TeslaFi Fleet Totals - 2,818 Active Cars - 2018.39.4 fe295e8 - 3 cars on that firmware among those 2,818.
So you could extrapolate that to the real car count and guess how many cars really have the update at best.
I am not sure what the number of total sold cars are. Maybe 350 000 ? ~0.8-1 % total of cars in TeslaFi "fleet". So if 10 cars are on some revision of v9 on teslafi then it might be around 1000 ? Who knows?
 
  • Like
Reactions: zoeconner
This an interesting tweet by Musk

"Tesla Autopilot team is awesome! Got to make sure we iron out the details though. Long tail of tricky edge cases. Important to get front wide angle, B pillar & side repeater camera neural nets just right."

Elon Musk on Twitter

I thought they weren't using the B Pillar cameras for V9.0?

Likely needed for lane change for coverage/ redundancy and to see if someone two lanes over is merging into the same spot. Motorcycle detection too?

Regarding lane change, can we go to voice commands? Perhaps gee and haw?
 
I never said that V9 was L3 because it could do lane changing. I said V9's ability to monitor the environment and to know WHEN to do a lane change brings it one step closer to L3.

*snip*

Oh boy here we go again.. time to grab some more popcorn!
Bill-Hader-Popcorn-reaction-Gif-On-The-Daily-Show.gif
 
Oh boy here we go again.. time to grab some more popcorn!
Bill-Hader-Popcorn-reaction-Gif-On-The-Daily-Show.gif

Yeah, I really did not intend to get dragged back into a debate about levels of autonomy. I was merely expressing my thoughts that V9 does bring autopilot closer to highway L3. But I accidentally misused a technical expression ("dynamic driving task") so I got busted by the "SAE police". LOL.
 
  • Like
Reactions: S100D and TT97
the human driver must be ready to take back control at any time when the ADS requests the human driver to do so

This is almost the only part of the L3 description that matters. All the rest of this is completely beside the point. An L2 is supervised by the driver. And L3 system supervises itself and tells the human driver when a takeover is required. The L3 system must allow some defined, specific period of time for the human driver to respond and during that time must continue to operate safely.

Nothing else really matters for L2 vs L3. As others have mentioned, you could have an L3 system not capable of lane changes at all. You could have a system which is L3 only when stopped, and it only needs to inform the driver to take over when being stopped is no longer appropriate (like maybe the car ahead started moving, or the light changed green) and it just has to stay stopped until the human takes over. Nobody would find much value in that system but it would be L3.

Nothing in anything we've seen from Tesla, not the hacked uninitiated lane change code, and not even the demo video actually, in which the human driver was clearly attentive and supervising, suggests that they have any system which is capable of reliably providing advance warning to the driver that she will need to take over. That is a fantastically, ridiculously hard problem and I honestly don't think Tesla has even started working on it in any serious way.