Tesla and Uber Unlikely to Survive...
Discussion
gangzoom said:
The in car display now shows traffic coming the other way, but its 1 second slow and misses 50% of the traffic.
The v3 hardware might reduce 1 second to half a second (doubling the speed would be a significant achievement in hardware these days). The new hardware however will not fix missing 50% of the traffic.In the garage situation, either the software is capable of extrapolating that there is a garage wall to miss, or it isn't. When the car is just beginning to move, it has all the time in the world to figure out it's surroundings. Bigger better hardware isn't going to change that.(*)
(*) OK that's a gross simplification, sufficiently more powerful hardware can enable algorithms that simply cannot be run on existing silicon, but v3 vs v3 hardware is not giving that sort of a step change as far as I can tell.
Tuna said:
In the garage situation, either the software is capable of extrapolating that there is a garage wall to miss, or it isn't. When the car is just beginning to move, it has all the time in the world to figure out it's surroundings. Bigger better hardware isn't going to change that.(*)
(*) OK that's a gross simplification, sufficiently more powerful hardware can enable algorithms that simply cannot be run on existing silicon, but v3 vs v3 hardware is not giving that sort of a step change as far as I can tell.
I don't disagree, but more sensors will simply send more data that cannot be processed. The real deal breaker is the software, which as far as I know even DeepMind hasn't cracked, so I doubt Tesla have.(*) OK that's a gross simplification, sufficiently more powerful hardware can enable algorithms that simply cannot be run on existing silicon, but v3 vs v3 hardware is not giving that sort of a step change as far as I can tell.
We have to wait and see what the v3 computer can do, from the point of view of tech its all fascinating.
jjwilde said:
I mean we will eventually get fully self driving cars. It's just a matter of debate about when.
Or does anyone here think it will never happen in our lifetimes (assuming we're all under like 40)?
I feel that’s pretty much right. While the tech still seems quite a long way off from even matching the abilities of a borderline retarded human the secondary hurdle of legislation also has to be considered. That will also take years and in the interim we may have company car schemes and private landlords stating that autonomous tech cannot be used etc. Even lease underwriters and insurance companies may make such stipulations as they own the assets and the liabilities. Or does anyone here think it will never happen in our lifetimes (assuming we're all under like 40)?
The whole solution is years away with many hurdles to work through but there is no reason why the end goal is not achievable and that the journey to that point won’t see limited uses being legalised where there are clear benefits, such as on more simplistic rule based roads such as motorways that also carry higher risk of mortality etc.
Bringing this back to Musk and Tesla, time isn’t on their side as it is for the tech giants or the automotive giants who have viable core businesses and big balance sheets etc and it is pretty clear that the ‘Million taxis by 2020’ was dishonest snake oil vending to whip up sales in the short term at a time when the company was in a very serious financial predicament and needing to get an emergency funding round banged out.
For me personally, the idea of self driving cars is hugely exciting and it’ll change the world we live in enormously. If only to get me to the pub and back when I’m old.
DonkeyApple said:
For me personally, the idea of self driving cars is hugely exciting and it’ll change the world we live in enormously. If only to get me to the pub and back when I’m old.
If your actually interested in the science than Nature is publishing more and more on AI, this latest article is fantastic. Any idiot can write anything on twitter/internet these days be promoted to instant glory, but a publication in Nature is still the gold standard for every single true scientist/researcher. https://www.nature.com/articles/d41586-019-03013-5
I still think DeepMind will crack true AI first, and probably explains why true 'speed sign recognition' in Tesla cars may never come, despite how easy a task it seems.
gangzoom said:
If your actually interested in the science than Nature is publishing more and more on AI, this latest article is fantastic. Any idiot can write anything on twitter/internet these days be promoted to instant glory, but a publication in Nature is still the gold standard for every single true scientist/researcher.
https://www.nature.com/articles/d41586-019-03013-5
I still think DeepMind will crack true AI first, and probably explains why true 'speed sign recognition' in Tesla cars may never come, despite how easy a task it seems.
I quoted some posts on that subject about nine months ago. Largely ignored https://www.nature.com/articles/d41586-019-03013-5
I still think DeepMind will crack true AI first, and probably explains why true 'speed sign recognition' in Tesla cars may never come, despite how easy a task it seems.
Unfortunately, rather like Fusion Reactors, AI has been 'a few years away' for decades now. To those new to the field it all seems bright and exciting, but I can quote you articles from thirty years ago saying much the same thing, and promising much the same breakthroughs 'any time now'.
Thanks.
Re sign recognition, when a car can travel down the last 300 yards of the A1 into London and can read and comprehend all 100 signs slung up there and then have the intelligence to understand that in the 30 years they’ve all been up there there has never been a single vehicle that has paid any heed to a single one of them and that the real rules of the road are an amalgamation of unwritten rules from a wide range of global cultures that are unique to that single junction then I will start to believe.
Re sign recognition, when a car can travel down the last 300 yards of the A1 into London and can read and comprehend all 100 signs slung up there and then have the intelligence to understand that in the 30 years they’ve all been up there there has never been a single vehicle that has paid any heed to a single one of them and that the real rules of the road are an amalgamation of unwritten rules from a wide range of global cultures that are unique to that single junction then I will start to believe.
Tuna said:
gangzoom said:
The in car display now shows traffic coming the other way, but its 1 second slow and misses 50% of the traffic.
The v3 hardware might reduce 1 second to half a second (doubling the speed would be a significant achievement in hardware these days). The new hardware however will not fix missing 50% of the traffic.Witchfinder said:
How quickly does a human typically react to seeing an object, or to a change in a situation?
One can imagine that a computer can react quicker than a human but that isn’t the issue at hand. The issue is which is better at identifying the object and understanding what it is and how it is going to interact with your environment. So far all we can safely say is that computers are better than humans who are asleep, reading a book, dead or heavily retarded etc. DonkeyApple said:
Witchfinder said:
How quickly does a human typically react to seeing an object, or to a change in a situation?
One can imagine that a computer can react quicker than a human but that isn’t the issue at hand. The issue is which is better at identifying the object and understanding what it is and how it is going to interact with your environment. So far all we can safely say is that computers are better than humans who are asleep, reading a book, dead or heavily retarded etc. coetzeeh said:
Not sure if this has been cnsidered previously but will virtual road signs become available for your on board processor to read?
Works in the world of commercial aviation.
Do you mean a radio transmitter? Or do you mean an up to date mapping system with the current speed limits included.Works in the world of commercial aviation.
Dave Hedgehog said:
hyphen said:
What an idiot you could see it totally had it wrong yet they kept going Witchfinder said:
hyphen said:
The firm, headed by inventor Sir James Dyson, said its engineers had developed a "fantastic electric car" but that it would not hit the roads because it was not "commercially viable".Gassing Station | EV and Alternative Fuels | Top of Page | What's New | My Stuff