-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Elon: "Feature complete for full self driving this year"
- Thread starter dckiwi
- Start date
I didn't say it's ok, I was just translating Elon-isms to real-isms.
mrbulk
Member
Don't know if this has been said yet (36 pages? Already??) but if the human error factor were taken out, then I'd be much more comfortable with "FSD", meaning if every car had the same FSD system...Hey, imagine a pre-planned Tesla City...
mrbulk
Member
So then why not use those as references instead of quotes from before the split? (Other than you would lose a year or two of material).
I'm sure you spent a lot of time and training to get your title, kudos.
Kant.Ing
Member
You can already do so (almost but maybe another software update), provided that your car uses only freeways.
mrbulk
Member
yeah now if he just had a solar panel on that tent...You can already do so (almost but maybe another software update), provided that your car uses only freeways.
View attachment 387865
Do you think they'll actually release NoA without confirmation in the next 6 months?
As it exists right now there are plenty of users like yourself or myself who feels like it has some serious issues.
You brought up a good example of how sometimes it seems to use proximity to an exit as to when to get over while failing to know which lane it's really supposed to be in for that exit. Other times the navigation is simply wrong, and has it taking an exit it shouldn't be taking.
Then we have the issues with the aborted lane changes that happens around 20% of the time (at least the last time I checked with the last 2018 release).
We've known for awhile that unconfirmed lane changes were being tested by a select group of people. So I fail to see what's different now. Aside from maybe a few more people on the EAP program. I haven't seen a whole lot leak out about the progress. So either people have gotten really good at not talking about fight club (which I doubt) or it's still a very tiny amount of select people. In that case it's likely months away.
The problem with releasing this before it's ready is the level of ballsiness that you brought up.
It's one thing to be ballsy, but it's another thing to be stupid.
Personally I think it would stupid to release it until the likelihood of it taking an exit it's not supposed to take is reduced to a very small number. If not it's going to result in drivers doing quick corrections from a lane that's ending to a lane left. Where they might do so in a rushed manner.
We can also pretty much guarantee that anything that lessens driver engagement, but maintains driver responsibility will result in a few crashes.
Even if the NoA was absolutely perfect in it's lane choices the human driver would intervene at times where the human was confused. Like "Oh, crap I think I wanted that exit..." followed by "Oh, sorry you were right".
Anyone watching the Tesla swerve around would be going "What the hell is that driver doing???"
I'd rather see this get put on the back burner, and instead for Tesla to focus on improving NoA along with navigation maps.
strangecosmos
Non-Member
Isn’t Alphabet (Google, DeepMind, Waymo) presumably better at deep RL than anyone else in the world? Including Mobileye?
Do you think Mobileye’s putative head start is enough to overcome Alphabet’s advantage in deep RL? Or do you think Alphabet just has no advantage over Mobileye in deep RL?
What do you think this will look like in practice? What will be the clear evidence that Mobileye has overtaken Waymo?
Does Intel have nearly as much compute as Google?
Including lidar? Isn’t any lidar you get that cheaply going to be low-range and low-resolution?
I definitely don’t think that with regard to perception. Obviously there was a huge advancement with deep supervised learning in 2012 and there’s been significant progress on this approach every year since.
With regard to action, my point is that hand coding software for autonomous cars is at least as old as the first DARPA Grand Challenge in 2004 (so, at least 15 years old). There’s been incremental progress in hand coding cars to drive, but it’s fundamentally the same approach today as people were using in 2004.
There have been fundamental advancements in action for cars and robots — now we have deep reinforcement learning, deep supervised learning of action (imitation learning), and inverse reinforcement learning (another form of imitation learning) — but not all companies apply (or can apply) these new machine learning approaches to the full scope of an autonomous car’s action.
I don’t know where you get “we haven't made an advancement in 15 years” from that, especially if you’re talking about perception.
Last edited:
It looks like it? I mean the news say so.
Yes!!! Enhanced summon is on its way.....
Tesla begins rollout of Enhanced Summon, Navigate on Autopilot with no stalk confirmation
Elon’s word still moves mountains amongst the teslarati and the hordes of newcomers.
As I said in the poll thread I believe, this one is very tricky to estimate and I haven’t really — because we know technically the feature is there and on the other hand we know driver will be responsible still — so it is really all about when and how Tesla has the guts to enable it and that could be at any time given their CEO who has a tendency for such moves.
A policy decision really, not about technology implementation at this stage. I doubt anyone is expecting any kind of gamechanger on technology level, we know what it is and any improvements are likely to be gradual only. So I do think we pretty much know what the feature is and will be like for the foreseeable future.
As @verygreen mentioned, they even have a remote switch for it now? Is that for enabling or disabling it is another question.
Last edited:
diplomat33
Average guy who loves autonomous vehicles
Well, according to reports the upcoming 2019.8 update has the code for NOA without confirmation, Tesla will just need to "flip a switch" to activate it. So that would suggest that we are close to NOA without confirmation. I think less than 6 months is very possible.
The safety concerns you raise is precisely why Tesla has waited so long to release it. Why do you think that Musk wrote that famous tweet about "10 million miles before Tesla will remove the confirmation" and why it's been 6 months after that tweet before we are even close to getting no confirmation? Precisely because it's a feature that has to work right and Tesla is taking their time as they should to get it right.
So yeah, I think it is a given that NOA will be much improved in other areas before we get no confirmation. I am eager to test NOA on 2019.8 when I get that update because I want to see if and how much NOA has improved. I think the level of improvement will be a tell tale sign of how close we are to getting NOA with no confirmation.
I think Tesla can walk and chew gum at the same time, as they say. Tesla can work to improve NOA along with nav maps AND work to get NOA ready for no confirmation at the same time. In fact, I would guess that is exactly what Tesla has been doing. They don't have to sacrifice one to do the other. In fact, I would argue that improving NOA along with nav maps is a prerequisite for removing the stalk confirmation. So they really need to do "all of the above". So it is almost required that the new NOA will be much improved in other areas too.
Because none of his statement were ever dependent on mobileye. How is it that you don't understand that what Tesla created (perception NNs) for their AP system they still had to create for their self driving system? All the things they have done so far today they still had to do if they still used mobileye. It wouldn't be any different. Nothing would have been faster. Their control algorithm wasn't changed, they already had 2 years of development already done for AP1 which they leveraged for AP2.
Absolutely nothing changes development wise with or without mobileye.
Bachelors in Computer Science.
4+ years of professional software engineering experience.
Currently working on Masters in Computer Science (Computer Vision/AI discipline).
Might do PHD in machine learning if i have free-time after that (probably wont because i think its a waste of time/money).
Not saying that to flex, said that to say that my point is, this is my career. I do this for a living. There's a huge difference between reading about a software methodology on Wikipedia versus actually doing it professionally. When you work in a company as an SE, you recognize the BS that Elon says is exactly that, BS.
May you have a long and successful career. But note, that when you start believing your own BS over historical facts, you can get into trouble. ( ancient Chinese proverb)
Not even Elon believes his own BS. When he get caught in a spot, he tries extremely hard to enable it and succeed. But it doesn’t stop him from reaching for high goals. With my PhD, artificial intelligence articles and R&D in the 1980s (In my day, we had rules...
), BS and wishful thinking are still common.Waymo, Cruise Automation, Uber, Audi and Tesla should deliver significant car automation this year. Personally I lust for the new computer upgrade, having always be a fan of faster hardware. The newly release of advanced summon could just be an modest update to using all the sensors in the existing Neural net.
I think of it as an initial development in this using all the sensors at the same time. It has to be at slow speeds eg parking lots, since the computer system needs more power to handle highway speeds.
Developing your own FSD code is easier if you are running in parallel with a known good data source to reference.
Ya don't say...
And here I though we might be contemporaries.
@mongo What... @Bladerskb ’s grammar didn’t give him away as a millennial to you?
All jest aside, Tesla’s AP progress with regards to Elon Musk’s comments really isn’t about software development in general. It is much more about the state of the art of autonomous vehicles and experience with that. (If even that, much of it is about viral PR.)
On that I’ve yet to see anyone else’s credentials I buy other than the one person that was driven away by moderator action. Whether or not someone knows generic software development is neither here nor there. Nor is 1980s AI.
All jest aside, Tesla’s AP progress with regards to Elon Musk’s comments really isn’t about software development in general. It is much more about the state of the art of autonomous vehicles and experience with that. (If even that, much of it is about viral PR.)
On that I’ve yet to see anyone else’s credentials I buy other than the one person that was driven away by moderator action. Whether or not someone knows generic software development is neither here nor there. Nor is 1980s AI.
Last edited:
@mongo What... @Bladerskb ’s grammar didn’t give him away as a millennial to you?
All jest aside, Tesla’s AP progress with regards to Elon Musk’s comments really isn’t about software development in general. It is much more about the state of the art of autonomous vehicles and experience with that. (If even that, much of it is about viral PR.)
On that I’ve yet to see anyone else’s credentials I buy other than the one person that was driven away by moderator action. Whether or not someone knows generic software development is neither here nor there. Nor is 1980s AI.
The mods killed someone?
Nah, I believe they just got annoyed and left. It is discussed here: Poll: should the Autonomous Vehicles forum have a moderator? But there was some screenshot evidence that was pretty compelling on their expertise.
That said I’ve felt @Bladerskb could well have worked in the self-driving industry. His PoV certainly is very different from the average software guy here in my view.
diplomat33
Average guy who loves autonomous vehicles
By the way, here's Elon Musk's quick bio:
Age: 48
Net Worth: $21.4 billion
Education:
Economics Degree from Queen's University
Studied Applied Physics and Material Science at the PhD level at the University of Pennsylvania (dropped out to become a billionaire).
Awards: Fellow of the Royal Society
Companies:
Founder, CEO and Lead Designer for SpaceX
Co-Founder, CEO and Product Architect for Tesla Inc.
Co-Founder and CEO of Neuralink
Founder of The Boring Company
Co-Founder and Co-Chairnan of OpenAI
Co-Founder of PayPal
That's quite the resume! He was also 21st on Forbes' List of Most Powerful People!
So yeah, I know it's fun to poke fun at Musk for his bad preidctions and you certainly have every right to disagree with him about FSD if you work in the self-driving field and you think your approach is better but the fact is that the guy is no slouch either. Anybody here start billion dollar companies that build electric cars, launch spacecrafts into space, or develop artificial intelligence? I think not.
Age: 48
Net Worth: $21.4 billion
Education:
Economics Degree from Queen's University
Studied Applied Physics and Material Science at the PhD level at the University of Pennsylvania (dropped out to become a billionaire).
Awards: Fellow of the Royal Society
Companies:
Founder, CEO and Lead Designer for SpaceX
Co-Founder, CEO and Product Architect for Tesla Inc.
Co-Founder and CEO of Neuralink
Founder of The Boring Company
Co-Founder and Co-Chairnan of OpenAI
Co-Founder of PayPal
That's quite the resume! He was also 21st on Forbes' List of Most Powerful People!
So yeah, I know it's fun to poke fun at Musk for his bad preidctions and you certainly have every right to disagree with him about FSD if you work in the self-driving field and you think your approach is better but the fact is that the guy is no slouch either. Anybody here start billion dollar companies that build electric cars, launch spacecrafts into space, or develop artificial intelligence? I think not.
Similar threads
