Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register
This site may earn commission on affiliate links.
He definitely said it and sadly I don’t think he was trolling. I do like how he admitted, he could do the CTC drive now, but so much would have to be hard coded for the route that it wouldn’t work for anything but that route and be to brittle. Aka, yah this isn’t solved, three months maybe, six months definitely
 
And that expalins specific coast to coast maps starting to pop into existence at the end of December how, I wonder? ;)

To be honest, I think they wanted to game a zero-disengagement drive using HD mapping and it didn't work out, so now they're pretending like they never wanted to do that in the first place :D
 
I wonder if Elon will release his tax returns.... what are you hiding Elon? Follow the money people... follow the money... frickin FSD is a sham and we are all being made into his sheep people.... sheeeepole. Oh my god, FSD was an inside job! Fluoride is turning the frogs gay... frickin Obama , what are we talking about again?
 
warning: someone hacked buttershrimp's account

You call that hacked. I call that buttershrimp Tuesday.

Have not been on this thread for a while and came here to post following link. So we're still just about to hit parity with ap1 based on test feedback?!?!!! Wtf. Fsd coast to coast had no hope for 3 months timeframe. Also, can they be any more underwhelming with this autopilot update?! Frustrated.

Tesla starts beta-testing new Autopilot update with new feature and more advanced neural net
 
  • Like
Reactions: 1 person
You call that hacked. I call that buttershrimp Tuesday.

Have not been on this thread for a while and came here to post following link. So we're still just about to hit parity with ap1 based on test feedback?!?!!! Wtf. Fsd coast to coast had no hope for 3 months timeframe. Also, can they be any more underwhelming with this autopilot update?! Frustrated.

Tesla starts beta-testing new Autopilot update with new feature and more advanced neural net

The trouble is, depending on who you ask, it either isn’t as good now, or was already better than AP1 back in November. When you get to a certain point, it’s hard to tell the difference between one system that rarely fails and another system that rarely fails, especially when the things they fail on are different. At this point, I just ignore comparisons against AP1 and am pleased that things will get better with my car.
 
  • Like
Reactions: chillaban
To be honest, I think they wanted to game a zero-disengagement drive using HD mapping and it didn't work out, so now they're pretending like they never wanted to do that in the first place :D

That's not really gaming it. That is how it's supposed to actually work. But I agree it didn't and gaming it would have meant specific code to deal with specific parts of the trip which they could have done but that would have been dumb, dishonest and wasteful. It's good they didn't. What Elon said on the call was interesting. Because he said what I thought I heard in the past, that after then demo the system should be ready soon after for release. Either it works or it's fake and doesn't work. So if it works, release it. I like that it seems to be what Elon is thinking.
 
Updated predictions on self-driving from Musk during the SXSW 2018 Q&A (3/11/2018):

"I think probably by end of next year [end of 2019] self-driving will encompass essentially all modes of driving and be at least 100% to 200% safer than a person. By the end of next year. We're talking maybe 18 months from now. NHTSA did studies on Tesla Autopilot 1 which is relatively primitive and found a 45% reduction in highway accidents. And that's despite Autopilot 1 being just version 1. Version 2 I think will be at least 2 or 3 times better. That's the current version that's running right now."

This was at about 3hrs, 48 min into the livestream:
 
  • Informative
Reactions: lunitiks
what a fraud and he had the nerve to say that other ppls batting average weren't good.
And yet again the host never reminded him of his failed predictions for the past 3 years.
This was just another opportunity for him to gainsay.
Elon did admit himself on a general basis his earlier time predictions have been quite a bit off, and that he is trying to recalibrate. It's almost at the start of the Q&A, you should see it all.
 
what a fraud and he had the nerve to say that other ppls batting average weren't good.
And yet again the host never reminded him of his failed predictions for the past 3 years.
This was just another opportunity for him to gainsay.

Wasn't a host, that was a friend of his.
Wasn't an interview, was a chance for people to ask questions they wanted to hear his response to.
 
I was completely blown away by a post over at reddit yesterday.

A guy called PM_YOUR_NIPS_PAPER - who says he is / was training Tesla's computer vision network - claimed:


I don't know if this guy's for real or not. Nevertheless, I find what he's saying a bit scary. Could it be true? What pieces of evidence or indications do we have either way?

A quick Google search tells me there are several software tools out there that could be used for object tracking / annotation. Even YouTube seems to have a tool for this.

YouTube-BB Dataset | Google Research
ViTBAT - Video annotation tool

Random video:

I really hope this isn't the case. Unfortunately that dude's post got me thinking ...

Oahi-lit-firebrand.jpg


Tesla Self-Driving Demonstration

some anonymous guy on reddit is not exactly what I would call a credible source.
 
  • Like
Reactions: DiamondHands
Obviously it is a preplanned route with a lot of hard coding. Afterall it is a future capability demo.

The visualizations are a representation of what the system infers. Very similar to the color images we see in astronomy of nebulae.

As long there is no human driving it in REAL TIME, sitting inside the car or even remotely it is not fake.

True but it would still be highly misleading and dishonest if true since folks watching the video would have every reason to believe they are seeing something different than what they are actually seeing.