Chipmaker Nvidia's CEO Sees Fully Autonomous Cars Within 4 Years (reuters.com) 77
An anonymous reader shares a report: Nvidia chief executive Jensen Huang said on Thursday artificial intelligence would enable fully automated cars within 4 years, but sought to tamp down expectations for a surge in demand for its chips from cryptocurrency miners. Nvidia came to prominence in the gaming industry for designing graphics-processing chips, but in recent years has been expanding into newer technologies including high-performance computing, artificial intelligence, and self-driving cars. Its expansion has been richly rewarded with a 170 percent stock surge over the past year, boosting its market value to $116 billion. "It will take no more than 4 years to have fully autonomous cars on the road. How long it takes for the vast majority of cars on the road to become that, it really just depends," Huang told media after a company event in Taipei.
CEO says X that will do Y (Score:5, Insightful)
X = something that may or may not happen in a few years.
Re: (Score:1)
Its one thing to 'have autonomous cars on the road', its another to have a product ready for mainstream. Programmer types tend to often underestimate engineering challenges of this magnitude.
Re: (Score:3)
So many of us have the attitude, "If I acknowledge and document a defect, it is a known issue. We will fix it when we can, if we want to, if it is important enough ..."
Oh, yeah, wait till you get up on the witness stand and the ambulance chaser asks you, "So, Ms Imac Oder, if your car is going at 79 mph, and you increment the set cruise speed by 1mph, the cruise control module will crash, and the throttle will open wide and the car will
Re: (Score:2)
So exactly where was Woz when he exceeded the speed limit by that much, and attempted to enter what was probably a "reckless driving" speed range?
Not in my lifetime (Score:1)
Way too many edge cases.
Re: (Score:3)
I'm not sure about "lifetime", but "any time soon"? Agreed. And I say this as a big Tesla fan. But the edge cases are just way too numerous, and not something you can just sweep on the rug with "oh, but it'll be safer because it never gets distracted...." Yeah, try telling that to the family of the person you just killed in a situation that a human could have easily avoided.
Re: Not in my lifetime (Score:1)
We need an internet superhighway for cars (Score:3)
We should try to create something like the internet superhighway but for cars :-)
Autonomous cars will be more successful if they can be partitioned away from the pesky human controlled cars. Creating special separted lanes for them would be a reasonable thing to do if we can assume that the flux in these lanes will rise or other benefits ensue (fewer accidents, more personal productivity leading to willingness to pay toll fees benefitting the highway system for everytone in return for not having to drive).
Re: (Score:2)
Move along, nothing to see here. (Score:5, Insightful)
Man selling autonomous car parts says big things about autonomous cars.
Re: (Score:2)
They are easier to steal?
Re: (Score:2)
Should autonomous cars mature in my lifetime, I'm going to laugh so hard when there's the first widespread autonomous car hack. Suddenly, a million vehicles start converging on a single point, blocking all traffic in the entire region...
Re: (Score:2)
/golfclap
Re: (Score:2)
Could be possible. /. even.
Considering how connected modern cars are already towards their manufacturers servers.
There where some simple lock hacks a few days/weeks ago on the internet, probably on
Yeah, right. (Score:5, Insightful)
Other than his cynicism in blatantly trying to raise his companies stock price self driving cars will never be fully automated until they have a good understand of human psychology as well as the rules of the road and laws of physics. Maybe driving in the nice wide roads and intersections in the US is relatively simple, but lets see these cars navigate a european or far eastern city where its very hard to get out of a side turn unless you push out, or streets that are 2 narrow for 2 way traffic and the automated car is coming down it but someone decides to come up the other way anyway.
And theres the true test - lets see one navigate itself around the l'arc de triomphe roundabout in Paris. Good luck with that Mr Huang!
Re: (Score:3)
Not even all roads in the US: big cities in rush hour, poorly marked rural roads in the rain, etc, etc.
Re: (Score:3)
Not even all roads in the US: big cities in rush hour, poorly marked rural roads in the rain, etc, etc.
Snow, black ice, obscured lines (mud, snow, etc.), obscured signs (bushes, etc,), broken traffic lights, lane closures and rerouting due to construction (i.e. lane shifts to the other side of the highway - my GPS hates this), and may other situations that are not really edge cases but part of everyday driving.
Re: (Score:1)
lane shifts to the other side of the highway
What????
Re:Yeah, right. (Score:4, Informative)
During construction or repair (for example of a bridge), when there are 2 lanes going each way with a median between them, they will sometimes merge down to 1 lane each way, then have a temporary road connecting them. The result is that the two lane separated highway becomes a 1 lane each way non-separated highway. From the point of view of GPS and maps, it looks like you have driven across the median and are now traveling in the wrong direction on the 2-lane highway.
I drew an ascii art diagram, but slashdot says I have too many 'junk' characters.
Re: (Score:2)
Oh, ok. I was thinking of some weirdness where that was the design.
Re: (Score:2)
https://en.wikipedia.org/wiki/... [wikipedia.org]
Re: (Score:2)
Yeah, exactly...
Re: (Score:2)
I'd like to see how an automated car handles exiting a grass parking lot after a concert or sporting event. If the car isn't willing to stick it's nose into harms way or quickly move to a different line, it will end up the very last to leave. And if it has been raining, will it be able to avoid getting stuck in the mud?
Re: (Score:2)
Seems to me that analysing the type of surface the car is on or is heading for or defects in the road is something else that automated car designers haven't yet bothered with.
Re: (Score:2)
Which of course can be fatal. Yet they're struggling on the "low hanging fruit" aspects.
Re: (Score:2)
I'd like to see how an automated car handles exiting a grass parking lot after a concert or sporting event. If the car isn't willing to stick it's nose into harms way or quickly move to a different line, it will end up the very last to leave.
This is the main thing that makes me hesitant about early self driving cars. I would like a car to handle rush hour traffic, but I don't want it to take an extra 10 minutes because it doesn't recognize it is in the slow lane.
Re:Yeah, right. (Score:5, Interesting)
Interestingly enough, "rush hour traffic" is what current Level 2 systems are best at. Not at picking lanes, but at everything else. Lots of cars to help them stay centred in the lane even when lane markers are bad, slow speeds, etc, etc. It doesn't mean you can ignore the road, but it does mean less having to constantly focus on a line of unmoving cars to avoid the situation where if you don't start moving as soon as the car ahead of you does, people behind you get mad and start honking.
Re: (Score:2)
The solution is simple... just get rid of all of the human drivers!
Re: (Score:2)
And theres the true test - lets see one navigate itself around the l'arc de triomphe roundabout in Paris.
Happens several hundred times a day. (*facepalm*)
All majour automobile vendors (BMW, Toyota, Audi, Mercedes) have self driving cars on test runs all over Europe since a decade.
Re: (Score:1)
[citation needed]
Re: (Score:2)
"Happens several hundred times a day. (*facepalm*)"
Really? You'll be able tp provide a link proving this then won't you.
Re: (Score:2)
You lack google foo?
Every majour car company is testing self driving cars since ages. Basically in every 'interesting' european city.
Re: (Score:2)
but lets see these cars navigate a european or far eastern city where its very hard to get out of a side turn unless you push out
I'd just like to see them navigate I-70 between Denver and Vail during the winter.
Re: (Score:2)
Well fortunately for the creators of self-driving cars you don't exactly get in-depth interviews with other drivers so very often human drivers go "WTF is he doing?" too. If something happens regularly by informal convention they'll presumably add it to the repertoire, otherwise they'll just be confused and stop. I have no doubt that they've already had to solve thousands of situations that weren't in the rule book.
Re: (Score:2)
They work well when people follow the rules. But this is France we're talking about. Rules are considered optional on the road.
Re: (Score:2)
OMG, the first time I drove in the UK I was doing work in Swindon... It was bad enough driving a stick for the first time on the "wrong" side of the road, I got to the "magic roundabout" in Swindon and had no idea how to get through it.
I basically tailgated another car figuring that if I rode on his bumper and followed him through the craziness I at least wouldn't get hit. It worked and I basically avoided that place for the rest of the trip...
Re: (Score:3)
lets see these cars navigate a european or far eastern city where its very hard
Why start with that when there are much easier use cases? To me, that seems kind of like many of the arguments I used to see against EVs: "I drive over 100 miles a day" or "I frequently take long trips". So? There are plenty of other use cases where they work great. Same thing with autonomous vehicles. They don't have to solve every edge case from day one, just the most common circumstances.
Most likely, the first deployment of fully autonomous vehicles on public roads will be in selected areas, perhaps geo
CEOs (Score:1)
Re: (Score:2)
I seem to recall Bill Gates predicting that TCP/IP would fail to become the dominant networking protocol. Boy was he wrong!! That was an epic failure of foresight.
And yet people still found it cheaper to buy a TCP stack for Windows 3.x from a third party than to switch operating systems.
Re: (Score:3, Insightful)
Pay? Everyone used Trumpet Winsock which was shareware though very few people paid.
https://tech.slashdot.org/stor... [slashdot.org]
And of course when Windows 95 came out it had a built in TCP/IP stack.
Re: (Score:2)
Pay? Everyone used Trumpet Winsock which was shareware though very few people paid.
There were several stacks, including software from TGV and Chameleon. Trumpet's was by far the worst-performing. It was OK for SLIP or PPP, but total garbage if you were doing Ethernet.
Re: (Score:1)
I only ever used it over a modem, and only really to pick up emails. Back then emails were plain text and probably only a few KB each. So Trumpet wasn't the bottleneck in that case.
Re: (Score:2)
For home users that's probably the dominant case. For business users the standard was an IPX network, usually Novell. Novell offered IP-in-IPX tunneling, so you didn't have to use an IP network to get internet access. Microsoft eventually offered their own TCP stack, which gave acceptable performance, but not great. TGV made the fastest stack. Cisco purchased them and had them developing an alternate stack for Windows 95, but they must have figured out that Microsoft was planning to eventually bring NT to t
Re: (Score:1)
Yeah, I remember when I got out of uni and started working there were still a few IPX and NBF [wikipedia.org] networks. You had Dos machines with a network redirector that allowed you to access networked file servers and network printers.
It was all kind of remarkable actually. Because MS didn't have a viable server OS NetBios was peer to peer. And getting network access to work inside a Dos interrupt handler must have been a nightmare. Bill Gates went crazy at the 64K low memory footprint and so Larry Osterman got it down
AD winter (Score:2)
We've already passed the hype peak. We're heading into the AD winter.
Re: (Score:2)
Definition of the phrase, please, as well as citations or research demonstrating your assertion, since car companies across the world are pouring huge amounts of money and time into autonomous driving technology.
Re: (Score:3)
Winter is the time of year when it gets cold, the days are shorter, work and recreation both slow down preceding the end of the year. It is usually used to describe something that is coming to an end of some sort or the end of one thing and the beginning of something new depending on the context.
In this case I would say he is leaving it open to interpretation either the money will dry up and the idea die or it will slow down and then evolve into something else. My guess is that we will end up with a very g
Re: (Score:2, Interesting)
Exactly. All hype curves are followed by a disillusion phase, when the technology (or whatever) is not living up to the hype. It is followed by another phase where the real applications are found.
It is now evident that good automated driving will take care of boring monotonous labor of driving for hours on highways. However, the city (and even rural) scenarios prove to be very, very difficult.
I've not seen a credible L3 (let alone L4 or L5) FuSa or SOTIF (let alone high-availability) concept at any of the v
Netcraft confirms it: human-driven cars are dying (Score:1)
Well he's dumb then (Score:1)
Re: (Score:1)
Self Delivery = death of Auto Dealers (Score:2)
In a related news... (Score:2)
Sounds vaguely familiar.... (Score:2)
Five years seems to a magic number....
https://www.pcworld.com/article/2010601/googles-brin-predicts-self-driving-cars-for-ordinary-people.html [pcworld.com]
Its Only Autonomous If.. (Score:2)
I can climb in the back seat and go to sleep, and tell it to wake me when we get there, or take the kids to school.
Its going to have to be full-up AI to do that - converse with it and you don't know for sure if its a machine or not.
I'm thinking 30 years or more... just a guess, tho.