-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
What the chances Tesla cars will be self driving in 3 years? Why do you think that way?
- Thread starter DanCar
- Start date
Completely separate question.
But in all three situations there were/are strong industry experts who looked at the "present" evidence (the limitations of the first roadster, the delays with the S and especially the X. The early Spacex Hopper experiments) and did/do not see anything to applaud.
Same with AP - just an equally unimpressive collection of hazardous experiments and prototypes which are a long way off being what is required, which ignore sensible industry practices.
We all remember the criticisms - hydrogen is / hybrids are the future / reusable rockets will never be cost effective.
Didn't make any difference.
Certainly Musk rubs people up the wrong way. Lots of people would dearly dearly love to see him fail to the point of obsession (paging @Bladerskb)
Doesn't mean it will happen.
The big difference is, that with Space-X and EVs, they were the only ones doing it. Autonomous cars were researched before Tesla even thought about it, by the likes of Google. It's not like Tesla were the only ones trying to build a self driving car, while the rest of the industry is looking at them and saying it's not possible.
They are all trying, but only Tesla(Elon) is giving out super optimistic targets, which they have never met so far.
Daniel in SD
(supervised)
Exactly. The EV1 came out over two decades ago. Was anyone really saying that it was not possible to build a mass market EV? I think the industry just felt it wouldn’t be profitable. It was never a technical question, just an economic one.
With autonomous vehicles you’ve got companies spending basically unlimited amounts of money to solve the problem because everyone knows that it would be incredibly profitable to sell a functional autonomous vehicle system. It’s not analogous at all in my opinion.
And who knows, maybe Apple will surprise everyone.
Or BYD will win via stolen IP
The point is about pioneering, disruptive research is that breakthroughs are unexpected and "unfair" for someone.
@Bladerskb believes that Amon is a brilliant engineer and a man of honour (both of which I have no reason to doubt) and that if there was any justice in the world, Amon would beat Musk to the prize.
Others see the level of investment from Alphabet and expect them to win instead.
None of these are a guarantee.
Musk is a difficult boss to work for and, by that measure, deserves to be bankwupt several times over, already.
Yet here we are.
Or BYD will win via stolen IP
The point is about pioneering, disruptive research is that breakthroughs are unexpected and "unfair" for someone.
@Bladerskb believes that Amon is a brilliant engineer and a man of honour (both of which I have no reason to doubt) and that if there was any justice in the world, Amon would beat Musk to the prize.
Others see the level of investment from Alphabet and expect them to win instead.
None of these are a guarantee.
Musk is a difficult boss to work for and, by that measure, deserves to be bankwupt several times over, already.
Yet here we are.
Not really.
Blue Origin was founded in 2000, Sierra Nevada dates from 1963, Orbital Sciences was 1982
Even John Carmack's Armadillo Aerospace was around in 2000
EV startups include AC Propulsion, Th!nk, and Aptera. Most were NEV "punishment vehicles" or quadracycles (with the notable exception of the T-Zero)
Can you give us more details about what you see happening? How well will it handle snow? How often will it false brake? How often will it not brake when it should? Can I sleep in the car while it drives me cross country? Will it handle driving through a recent crash site? Will it handle a policeman giving hand signals? Will it handle San Fran double parked cars requiring crossing of double yellow lines?
FSD is not possible with current hardware. AP/AS gets blind with little snow buildup. There are no washers for camera, they can blocked up easily.
It really depends on the definition. If the definition is 100% in every conceivable situation, then no, that won't happen--it may never happen. That is an unrealistic expectation. If you mean for highways and most in-town situations in good weather, then yes, there's a very good chance, probably close to 100%. If it's risky for a human driver, it will be risky for AI.
The prime virtues of AI driving are that AI is always attentive and doesn't drowse off. Most accidents that aren't the result of alcohol or just plain being stupid (e.g. street racing), are from a moment's inattention (distraction or blinded by the sun), falling asleep at the wheel, a medical condition, or making the wrong choice in an emergency situation (training). These are the things that AI is very good at preventing. I don't expect AI to get me through an ice day in DFW. Really I shouldn't be driving then either but it's show up at work or lose the job, so I go anyway. The rain/mud on the sensors is a solvable problem for a later version of the hardware. Let's get it working in good weather first.
The prime virtues of AI driving are that AI is always attentive and doesn't drowse off. Most accidents that aren't the result of alcohol or just plain being stupid (e.g. street racing), are from a moment's inattention (distraction or blinded by the sun), falling asleep at the wheel, a medical condition, or making the wrong choice in an emergency situation (training). These are the things that AI is very good at preventing. I don't expect AI to get me through an ice day in DFW. Really I shouldn't be driving then either but it's show up at work or lose the job, so I go anyway. The rain/mud on the sensors is a solvable problem for a later version of the hardware. Let's get it working in good weather first.
M3BlueGeorgia
Active Member
Perhaps, but not in all locations in all conditions.
But within in three years we could reasonable expect to see most locations in most weather conditions.
There's a lot of really gnarly edge cases to be solved before you get to "almost all locations" in "almost all weather".
Then the leap from that to all locations in all weathers is going to be almost unimaginably hard.
diplomat33
Average guy who loves autonomous vehicles
I could be wrong but I think "all weather" and "all locations" would be L4 or even L5 autonomy. L3 autonomy means the driver does not need to pay attention in certain situations but the car will notify the driver when they need to pay attention again. So I think L3 could work for all locations and all weather because the car would notify the driver when it can't handle something.
It is already legal in Arizona. There is no law that prohibits full self driving cars.
CarlK
Active Member
Elon said in the podcast "we have just rolled out the AI chip three years in the making". They had the right answer at least three years ago. During those three years people are busy second guessing without even having the slightest idea of what were going on in there. I don't think we still have slightest idea of what are going on there other than those that are already revealed to us.
It certainly could but I’m not sure people grasp the enormity of this: It is a much bigger leap from Level 2 to urban Level 3... than it is from urban Level 3 to urban Level 4, possibly even Level 5.
Urban Level 3 is a very, very tough nut to crack and not dissmilar to urban Level 4 at all. What mostly makes Level 3 more plausible (but still hard) compared to Level 4 is that it can be introduced first for the highway which is easier (in fact a highway Level 4 is also possible but that’s digressing). But go to the urban scenario, making a Level 3 car capable of driving there is immensely hard.
This is why it is so hard to see Tesla’s way out from Level 2 at this time. Level 3 requires car responsible driving. Tesla’s sensor suite redundancy is so low and their progress so far so sporadic that the move to car responsible driving seems a stretch even on the highway let alone on city streets.
Urban Level 3 must be able to handle everything, all obstacle debris types, all traffic signs, children running into the street at any moment, bicycles, slow-speed maneuvering in close corners etc. and everything without the driver because the driver must be given ample warning before getting his/her help...
It was not a secret Tesla was making a NN accelerator chip. It was discussed all over internet for years.
What was unexpected was how bad Tesla turned out to be with their NN progress since 2016...
CarlK
Active Member
Yeah right. You never agreed that's a significant factor. Remember all those brain vs vision debate we were having here?
The NN chip?
I do believe the NN chip is a significant factor in enabling Tesla’s progress as their approach does require quite a bit of compute. It is quite possible — aside from alienating MobilEye (and the questionable decision to not add more sensor redundancy of any kind when moving to AP2) — that on the hardware front Tesla’s moves have been pretty good here.
I do think it is widely agreed on the software front Tesla had little idea what they were doing prior to enlisting Karpathy and the biggest question indeed is how and when can Tesla make the software work, much less a question of can they ship the hardware...
CarlK
Active Member
Well you're at it again. All those wild guesses without the slightest understanding of what are going on there. If even you can think of it's better to do this or add that what's the chance you think the Tesla team had not considered that?
Similar threads
- Poll
