Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Nvidia on new self-driving system: “basically 5 years ahead and coming in 2017”

electrek.co

Post reply on HN
Print view View original

61–70 of 174 posts

Re: Nvidia on new self-driving system: “basically 5 years ahead and coming in 2017”

#61

Post by vic20forever »

sp332 wrote: post #3

Who thought making decent autonomous cars was going to take as long as 2020? I mean for regulations etc maybe, but not for the tech. Earlier this year I expected it to be done later this year. Edited for clarity (I hope).

One thing I hope the engineers are addressing: Google Maps recently directed me down a cobblestone street that required me to slow to a crawl. If an autonomous system attempted to maintain the speed limit on that road, I'm not sure the car would make it without anything being broken.

Re: Nvidia on new self-driving system: “basically 5 years ahead and coming in 2017”

#62

Post by arkitaip »

Retric wrote: post #16
tener wrote: post #10

Very cool: with capable CUDA device in a car, one should be able to mine some crypto coins with it! With some luck the car should be able to pay for itself! Alternatively, having in mind Elon's creative approach to finances, Tesla could get some significant hashing power with its fleet! Edit: looking at the downvote I suppose the joke wasn't obvious?

In the 9years I have been on HN I have up voted 3 really good jokes and downvoted thousands of so so ones. Before you post a joke think is it that good. Because, jokes are like spam on message boards they add clutter without aiding the discussion.

HN is a humor free zone and I for one appreciate it!

Re: Nvidia on new self-driving system: “basically 5 years ahead and coming in 2017”

#63

Post by ChuckMcM »

mattnewton wrote: post #58
ChuckMcM wrote: post #45

Earlier quoted context omitted.

The argument reasoning I've heard goes like this; People drive reasonably well using vision primarily and with imperfect visibility of their environment. Computer learning networks can classify imagery at least as accurately as humans and sometimes more so. A computer using imagery that is well classified from an array of visual sensors with near perfect visibility should be able to drive as well, or better, than a h…

Why not shoot for better than human performance and "cheat" any way possible along the way? To paraphrase a quote I can't remember by who, do we care if a submarine "swims"? Besides, with even with lidar, the problem is hard enough.

I agree, being better than human is a sales point that I expect to see in brochures. One of the ways I would expect that plays out is self driving transport cars for high value targets like world leaders and drug lords. "This car will respond faster, and more accurately, to get you to safety before a human driver even knew there was a problem."

That said, John stated that without LIDAR you couldn't adequately meet the environmental challenges and achieve good self driving.

Specifically "I'm still not happy with self-driving on vision alone ... There are too many hard cases for vision." which boils down to a disbelief on the imaging processing side of the pipeline where NVidia has been attacking using GPU type architectures to extract image information rather than generate it.

One way to evaluate how far the image processing pipeline has come is to look at research on how well it can classify images. And in that space, in the research, it is doing better than humans [1]. As I've said elsewhere I think LIDAR was a crutch that worked well to cover for weaknesses in classifying images, but I recognize that the crutch may no longer be needed (certainly Tesla and Nvidia are trying to make that case).

[1] http://www.eetimes.com/document.asp?doc_id=1325712

Re: Nvidia on new self-driving system: “basically 5 years ahead and coming in 2017”

#64

Post by vidarh »

tedunangst wrote: post #50
ChuckMcM wrote: post #45

Earlier quoted context omitted.

The argument reasoning I've heard goes like this; People drive reasonably well using vision primarily and with imperfect visibility of their environment. Computer learning networks can classify imagery at least as accurately as humans and sometimes more so. A computer using imagery that is well classified from an array of visual sensors with near perfect visibility should be able to drive as well, or better, than a h…

ML seems pretty bad at classifying things it hasn't seen before though. There are quite a few examples where an input outside the training data resulted in misclassification. Humans may not always see a white truck in a snowstorm, but is computer vision going to see it either? Or will it pattern match the few visible parts as something else entirely? Or dismiss the truck entirely as noise?

But the computer vision systems can be endlessly improved and merge experience from millions of cars, while human drivers accumulate experience from a single driver, age, and are eventually replaced by younger, inexperienced drivers.

Soon enough these systems will have data from encounters with far more varied situations than any single human will ever be physically able to encounter in a lifetime.

Re: Nvidia on new self-driving system: “basically 5 years ahead and coming in 2017”

#65

Post by Shivetya »

ChuckMcM wrote: post #45
Animats wrote: post #20

I'm still not happy with self-driving on vision alone, or vision augmented with radar. There are too many hard cases for vision. Everybody who has good self-driving right now - Google, Otto, Volvo, GM - uses LIDAR. Self-driving is coming to the first end users in 2017, in Volvo's test of 100 vehicles. Volvo has multiple LIDARs, multiple radars, multiple cameras, redundant computers, and redundant actuators. They're b…

The argument reasoning I've heard goes like this; People drive reasonably well using vision primarily and with imperfect visibility of their environment. Computer learning networks can classify imagery at least as accurately as humans and sometimes more so. A computer using imagery that is well classified from an array of visual sensors with near perfect visibility should be able to drive as well, or better, than a h…

LIDAR is just another form of seeing, just not as we are used to as people but combined with cameras they two would compliment each other. Relying on only one is a fool's gambit.

LIDAR won't go blind from white trucks on sunny days. LIDAR won't suffer snow blindness or inability to track in conditions where humans don't see well, like heavy rain at night. You add in visual acquisition to fine tune what you are detecting if necessary; perhaps to read signs and tell what color the traffic light is, maybe even to see brake lights. To know a floating bag is just that and not a solid object, to see that road is washed out or such.

Re: Nvidia on new self-driving system: “basically 5 years ahead and coming in 2017”

#66

Post by seanmcdirmid »

kchoudhu wrote: post #60
TaylorAlexander wrote: post #54

Earlier quoted context omitted.

Humans don't use accurate depth maps to drive. Machine learning is great at complex algorithms if done well.

Do the newer systems assume network connectivity to a backend processing facility? Personally, I view each additional layer (network, someone else's data center, machine learning) to be something that can fail and put people in danger. Assurance via local brute force is much more reassuring for me.

I can't see a system where processing is done remotely as working very well, nor is it very necessary as the hardware requirements shouldn't be that high. Training could definitely occur in a data center, but that is done offline.

Re: Nvidia on new self-driving system: “basically 5 years ahead and coming in 2017”

#67

Post by petra »

ghaff wrote: post #8
sp332 wrote: post #3

Who thought making decent autonomous cars was going to take as long as 2020? I mean for regulations etc maybe, but not for the tech. Earlier this year I expected it to be done later this year. Edited for clarity (I hope).

Many people. And I still don't expect to see general-purpose (secondary roads/cities, range of weather conditions) for decades.

Why would secondary cities be an issue ?

They seem big enough to incentivize mapping costs.

Re: Nvidia on new self-driving system: “basically 5 years ahead and coming in 2017”

#68

Post by Razengan »

greglindahl wrote: post #40
Razengan wrote: post #18

Earlier quoted context omitted.

The idea of ubiquitous and powerful computers in eveyrone's possession contributing to some global processing while they're idle, is actually fascinating.

It was, until most of those computers got a power-saving mode -- electricity is not free. Also, note that some things (like bitcoin mining) are so specialized that they run poorly on general-purpose cpus and gpus. Still, there is a thriving set of "do this computation at home" projects: Folding@Home, SETI@Home, ...

Well now billions of people have phones, tablets and laptops combined, so each of them can contribute some negligible use of their CPU while they're plugged in and sleeping, and they'll still add up to a huge difference to various projects.

I think something like this should be built into iOS/Android. You choose a single project to contribute to while your devices are idle, and you get some standardized currency/points based on how much computing you've donated.

Re: Nvidia on new self-driving system: “basically 5 years ahead and coming in 2017”

#69

Post by matheweis »

I thought the reason for Tesla switching away from Mobileeye was that Mobileeye and Tesla couldn't come to an agreement on price and data licensing?

https://electrek.co/2016/09/15/tesla-vision-mobileye-tesla-a...

... and because Mobileeye wasn't comfortable with Tesla using their system for level 4 & 5 driving:

https://electrek.co/2016/09/16/mobileye-responds-to-tesla-ag...

Re: Nvidia on new self-driving system: “basically 5 years ahead and coming in 2017”

#70

Post by ghaff »

petra wrote: post #67
ghaff wrote: post #8

Earlier quoted context omitted.

Many people. And I still don't expect to see general-purpose (secondary roads/cities, range of weather conditions) for decades.

Why would secondary cities be an issue ? They seem big enough to incentivize mapping costs.

I was referring to secondary roads (not limited access) and cities generally. Sorry I was unclear. Major cities would actually seem to be the bigger challenge than smaller ones in general.

Detailed and current mapping will likely be a necessary part of fully-autonomous systems at least initially. But there's nothing especially difficult about doing that mapping. It's just a question of economics.

Post reply on HN