Enhanced Summon coming (Elon tweet 6 Apr, 2019)

[Knightshade]

Sort of like a low-power, low quality, low resolution radar. Which works at greater distances and at highway speeds. Again, if you've cracked this nut for parking lots, you've solved FSD.

Again, not really.

Since the radar isn't 360, but the ultrasonics are. And they're much more useful at 1-5 mph than 65 mph.

There's certainly some overlap between these functions, but "will travel up to 150 feet to a fixed spot in a parking lot at 5 mph" is absolutely not "solving FSD" on every type of public road at 60-90 mph.

 

[⚡️ELECTROMAN⚡️]

Our cars are all full self driving. We’re all driving them ourselves. My self drove my car today fully by myself.

Elon wins \0/

 

[DrDabbles]

Our cars are all full self driving. We’re all driving them ourselves. My self drove my car today fully by myself.

Yes, we've seen this joke in this thread several times now. We get it.

 

[PRLM3]

Well said.

Totally agree. Elon vs the legal department - now that's a smackdown! Also agree that a) it will likely be close to Christmas (?Q1 2020) before we see anything real, b) what we see will be "watered down" substantially / phased in, c) worst job on the planet working on that development team! Read a couple stories of Elon replacing entire team a couple times including the lead. Sounds like there were even philosophical differences over basic approaches to the problem. They're trying trying to crack an extraordinarily difficult problem; hats off to them. Fun to watch from the outside though and I'm still bullish on one day getting into my 3, setting my destination, then firing up my laptop while my wife loads Netflix on the nav screen, and the next time I look up my destination is coming into view ...

 

[⚡️ELECTROMAN⚡️]

Yes, we've seen this joke in this thread several times now. We get it.

I come up with some pretty good stuff after a few drinks. At least it seems like it at the time.

 

[mazers]

"Extensive testing". As in random people with zero testing protocol, almost all of whom have zero testing experience, trying to make their cars poorly navigate public spaces, enraging other cars that get blocked by slow moving or stopped driverless vehicles. Not a good testing plan.

You don't think that Tesla has people putting the software through extensive, standard testing procedures also? EAP members who have this software release are simply providing some real life scenarios. The play with it sometimes, showing friends, and at other times use it in real situations like coming to a building's entrance when it is raining. No way to tell how many have it, so it may be a large number.

 

[Octo]

You don't think that Tesla has people putting the software through extensive, standard testing procedures also?

Do they? Why do they not fix obvious bugs and keep releasing updates without fixing them for months?

 

[DrDabbles]

You don't think that Tesla has people putting the software through extensive, standard testing procedures also?

Software regressions, maybe? But we don't see a ton of evidence for that. Behavior regressions? Not at all, no.

EAP members who have this software release are simply providing some real life scenarios.

Most of the scenarios we see on youtube are fairly dissimilar to real world testing. And in either case, without some formalized method to report behavior given scenarios, including detailed descriptions of conditions and patterns, then their scenarios aren't helpful. Training AI systems isn't just saying "it didn't work".

The play with it sometimes, showing friends, and at other times use it in real situations like coming to a building's entrance when it is raining. No way to tell how many have it, so it may be a large number.

It's not a representative sample, so the number isn't really the important thing here. The crux of the problem is that the development testing is obviously happening in the environment around Tesla, which is not only not representative of the rest of the planet, but is likely contributing to a false sense of success. Something Tesla seems to have been exhibiting for a while now. They've made claims for years that the AP system works significantly better than public evidence seems to suggest.

 

[mazers]

Software regressions, maybe? But we don't see a ton of evidence for that. Behavior regressions? Not at all, no.



Most of the scenarios we see on youtube are fairly dissimilar to real world testing. And in either case, without some formalized method to report behavior given scenarios, including detailed descriptions of conditions and patterns, then their scenarios aren't helpful. Training AI systems isn't just saying "it didn't work".



It's not a representative sample, so the number isn't really the important thing here. The crux of the problem is that the development testing is obviously happening in the environment around Tesla, which is not only not representative of the rest of the planet, but is likely contributing to a false sense of success. Something Tesla seems to have been exhibiting for a while now. They've made claims for years that the AP system works significantly better than public evidence seems to suggest.

I'd say that the people posting youtube videos is a very small minority of those testing Enhanced Summon. EAP users are under a non-disclosure agreement and agree not to post videos and to not have their software levels reported by Teslafi or other apps. Trust me, Enhanced Summon is getting tested in real world scenarios with feedback sent to Tesla.

 

[JulienW]

I'd say that the people posting youtube videos is a very small minority of those testing Enhanced Summon. EAP users are under a non-disclosure agreement and agree not to post videos and to not have their software levels reported by Teslafi or other apps. Trust me, Enhanced Summon is getting tested in real world scenarios with feedback sent to Tesla.

Also the people posting are "borrowing" an unidentified person's/3's for the videos.

 

[muleferg]

Still Waiting

 

[lolder]

Dr. Who is letting me use his time machine so I can get the "Free Beer Tomorrow" today.

 

[M109Rider]

Do they? Why do they not fix obvious bugs and keep releasing updates without fixing them for months?

Omg. You need to answer that question yourself.
Think hard, ponder the scrnario’s, you’ll get there. ..

 

[Octo]

Omg. You need to answer that question yourself.
Think hard, ponder the scrnario’s, you’ll get there. ..

What’s a scrnario?

 

[M109Rider]

What’s a scrnario?

Lol.
Well, I see you caught and figured out there’s a spelling error at play.
Surely you can figure the rest out. :

 

[Octo]

Lol.
Well, I see you caught and figured out there’s a spelling error at play.
Surely you can figure the rest out. :

Oh, I just wanted to be sure what I should ponder!

 

[nvx1977]

every time I see this thread I get excited that the update is imminent. then I see the date of the OP

 

[jebinc]

every time I see this thread I get excited that the update is imminent. then I see the date of the OP

At some point, the OP’s thread title will be accurate.

 

[lukex4]

At some point, the OP’s thread title will be accurate.

Yep, next week!

 

[ZoomsansVroom]

For whatever the value of my opinion is, I think it would be interesting if Tesla came out with a "This is really hard, so we've developed these line of reflectors/symbols which can be read to guide a car through a parking structure, without any people allowed in it", and then licensed that product to other manufacturers..." (I was "almost" going to be an architect, once, and I like what that would open up for commercial/multifamily building design... Of course, I also think Tesla should look at licensing motor and battery design to power tool manufacturers, so...)