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

Inside Waymo's Secret World for Training Self-Driving Cars

theatlantic.com

Post reply on HN
Print view View original

131–140 of 146 posts

Re: Inside Waymo's Secret World for Training Self-Driving Cars

#131

Post by bluGill »

jpindar wrote: post #127
jvolkman wrote: post #10

Earlier quoted context omitted.

And Waymo has already said that its cars will safely pull over when the weather gets too rough.

So during each winter storm, the roads will be lined with cars full of people freezing to death (literally) because their cars decided they can't drive home. And people trying to walk (in the roadway) to shelter, which isn't much safer.

Most cars can stay warm for several days just running the engine at idle. Assuming you get enough fresh air (not always a given in a snow storm that can block air vents) stopping and just running the engine for heat is safe unless your fuel tank is low - which it shouldn't be.

Re: Inside Waymo's Secret World for Training Self-Driving Cars

#132

Post by bluGill »

skywhopper wrote: post #97

I'm disappointed the article wasn't a bit more skeptical of some of the claims. Certainly the simulation-based testing is a good thing, but stats about how many billions of simulated miles have been driven can create a self-reinforcing delusion if everyone involved isn't careful to remember that the simulations can only work with well-known and expected situations. It sounds like Waymo realizes this and is building a…

One millions miles of different situations is worth more than 100 miles anywhere. Even your most carefully chosen worst case route won't cover all situations.

To get close you have to pay people to send their kid running between parked cars into the street at just the right moment. Nobody sane will put a real kids into that situation, but the simulation can (should!) have thousands of different variations. The simulation can afford to run all of those situations regularly. Simulation can ensure it is raining when you need it.

You need miles of real driving too, but they are too expensive and unpredictable to rely on for more than a final check.

Re: Inside Waymo's Secret World for Training Self-Driving Cars

#133

Post by Retric »

notahacker wrote: post #126
Retric wrote: post #116

Earlier quoted context omitted.

Can you give an actual example? Because, not being able to identify something is not necessarily an issue as long as the car notices something is there and it should not hit it. EX: I am sure the car had no idea what this was: https://youtu.be/Uj-rK8V-rik?t=26m11s but as long as it can tell it's bigger than a bread box and so it should not to hit it that's enough.

The obvious problem scenario is when unpredictable evasive action puts an autonomous vehicle into the path of other drivers (with human response times). That's when very sharp responses to an uncategorised "obstacle" that's actually a drifting plastic bag or a reflection cause more problems than they solve. Similarly, instantaneous harsh braking might help an AI save the small child it didn't anticipate might chase t…

A bag* or child running into a street is not an usual event, also a car is not going to 'evade' into another car. Rare events are the things people don't see across multiple human lifetimes not just something you don't see every month.

Which IMO is what's missing from the debate, unusual events are in terms of ~10+ million miles of training data before these things are in production. They are clearly out there, but I doubt people are going to react well to say someone falling from an overpass onto the road very well either. So, it's that narrow band of really odd but something a person would respond correctly to that's the 'problem'.

PS: Of course the bag might relate to a bug which are likely. But, IMO that's a completely different topic.

Re: Inside Waymo's Secret World for Training Self-Driving Cars

#134

Post by ghaff »

nradov wrote: post #124
sangnoir wrote: post #122

Earlier quoted context omitted.

> I have no speech impediments and speak US English with a neutral accent FYI: there is no such thing as a "neutral accent" - unless you mean neutral to your locale. I could describe an object as having "neutral temperature", but you'd need to know how hot/cold its environment is before it makes sense.

That's not correct for US English which absolutely does have a neutral accent unrelated to locale. For typical examples listen to nationwide network news broadcasts.

Well, yes, because an accent generally associated with the Northeast and Midwest became "Network Standard." The Wikipedia article on the topic is quite good. https://en.wikipedia.org/wiki/General_American

It's true that this accent generally lacks the strong regionalisms that some other US accents do. But it's mostly still an outgrowth of a general region even if its use is cultivated more widely.

Re: Inside Waymo's Secret World for Training Self-Driving Cars

#135

Post by bullfightonmars »

skywhopper wrote: post #97

I'm disappointed the article wasn't a bit more skeptical of some of the claims. Certainly the simulation-based testing is a good thing, but stats about how many billions of simulated miles have been driven can create a self-reinforcing delusion if everyone involved isn't careful to remember that the simulations can only work with well-known and expected situations. It sounds like Waymo realizes this and is building a…

> The story about being flummoxed by multilane roundabouts is depressing

Don't underestimate the bizarre engineering of infrastructure in Texas. Every road and intersection is bespoke, unlike any other you have previously experienced. It is quite difficult to navigate Austin as a human driver, I suspect it makes for a good testing ground for Waymo.

Re: Inside Waymo's Secret World for Training Self-Driving Cars

#136

Post by euyyn »

philjohn wrote: post #66

Unless I didn't understand correctly, on the animated image of the car turning, which included a wireframe representation of the scene, as well as the cameras: https://cdn.theatlantic.com/assets/media/img/posts/2017/08/W... The generated geometry doesn't seem to include the Bike that quickly passes behind the car at the intersection ...

The bike is the red box that crosses the intersection. It doesn't pass behind the car, but by its side. The white wireframe box that "detaches" from the Waymo car is where the cameras are (so what the car actually did). I assume the Waymo car that doesn't stop in the animation is what it should have done instead.

Re: Inside Waymo's Secret World for Training Self-Driving Cars

#137

Post by jvolkman »

euyyn wrote: post #136
philjohn wrote: post #66

Unless I didn't understand correctly, on the animated image of the car turning, which included a wireframe representation of the scene, as well as the cameras: https://cdn.theatlantic.com/assets/media/img/posts/2017/08/W... The generated geometry doesn't seem to include the Bike that quickly passes behind the car at the intersection ...

The bike is the red box that crosses the intersection. It doesn't pass behind the car, but by its side. The white wireframe box that "detaches" from the Waymo car is where the cameras are (so what the car actually did). I assume the Waymo car that doesn't stop in the animation is what it should have done instead.

More specifically, the article indicates that the car that doesn't stop is what the software - after changes to fix the original problem - would do now if it encountered the same situation.

Re: Inside Waymo's Secret World for Training Self-Driving Cars

#138

Post by Aron »

nopinsight wrote: post #19

The fact that Waymo revealed their "secret" tools for advancing this crucial technology implies that either: 1) They believe no one can quite catch up before they can launch the technology. Since they know several competitors have huge resources and brilliant people, it means they are quite close to launch. 2) These tools have become open secret within the industry, so no harm is done to their competitive position by…

The rate at which high level engineers move around leads me to believe that quite a few companies will arrive at approximately the same level of autonomy at the same time. Thus the marketshare for autonomy providers is probably going to be based on other factors than the actual technical superiority. Waymo will presumably need buyers, many of whom are the type that like IBM Watson style marketing.

Re: Inside Waymo's Secret World for Training Self-Driving Cars

#139

Post by Animats »

ghaff wrote: post #118
Animats wrote: post #46

There's a lot of misunderstanding about self-driving. Mostly because nobody is publishing much. If you want to do it right, you start with geometry. The first step is capturing range imagery and grinding it down to a 3D model of the world. This tells you where you physically can go. That's where we were at the DARPA Grand Challenge over a decade ago. Then comes moving object popout. What out there isn't a stationary…

One of the things that seems to be going on is that, after relatively limited visible progress for a long time, the ability to collect and analyze large datasets suddenly produced fairly striking results. And this in turn has led to a lot of thinking along the lines of "We just need to collect more data and crunch the numbers a bit better." As you suggest, for at least some classes of problems, we'll find that this a…

That's what seems to be happening in the various image classification competitions. Accuracy has reached 95-96%, but further improvement is tough.

Re: Inside Waymo's Secret World for Training Self-Driving Cars

#140

Post by notahacker »

Retric wrote: post #133
notahacker wrote: post #126

Earlier quoted context omitted.

The obvious problem scenario is when unpredictable evasive action puts an autonomous vehicle into the path of other drivers (with human response times). That's when very sharp responses to an uncategorised "obstacle" that's actually a drifting plastic bag or a reflection cause more problems than they solve. Similarly, instantaneous harsh braking might help an AI save the small child it didn't anticipate might chase t…

A bag* or child running into a street is not an usual event, also a car is not going to 'evade' into another car. Rare events are the things people don't see across multiple human lifetimes not just something you don't see every month. Which IMO is what's missing from the debate, unusual events are in terms of ~10+ million miles of training data before these things are in production. They are clearly out there, but I…

> A bag* or child running into a street is not an usual event, also a car is not going to 'evade' into another car

Opting to drive around a (stationary, visible from a distance) bag in an unpredictable manner is literally how Waymo's first "at fault" accident occurred...

The point is that a human has a concept of a "ball" linked to the concept of "children play football" and an understanding that if one sees the former, one should be prepared for the latter to bursts onto the road from behind the partially-obscured roadside. Appropriate action probably involves easing off the accelerator and lightly tapping the brake so the car behind gets a hint that you might have to stop suddenly on if a child emerges from behind a bush. An autonomous car which fails to anticipate even though it's lightning fast at slamming the brakes on is going to get rear ended a lot more.

The neural network of a self driving car might be able to classify small coloured spheres in the vicinity of the roadway as balls, and the AI will certainly have been taught the concept of a human-shaped obstacle moving across the roadway being a "need to stop" situation, but is unlikely to "learn" the association between the two through a few tens of million miles of regular driving, because only a very small proportion of "need to stop" events involve balls (and only a very small number of sightings of spheres moving in the vicinity of the roadway result in "need to stop" events). Of course, you can hard code a machine to respond to ball-shaped objects moving near roads by slowing down and you can construct a huge number of artificial test scenarios involving balls to teach the AI the association between balls and small children, but either of these options involves engineers envisaging the low frequency hazard and teaching it enough permutations of the sensory input for that hazard for it to be able to anticipate it (and there's a balance to be struck, because nobody wants a paranoid AI which drives through the city braking every time it sees something its neural network identifies as a pedestrian or the front of a parked car protruding from a driveway) Suffice to say, we take for granted our ability to know how to react to things like children chasing balls, staggering 4am drunks, tiny puddles the car in front just drove through versus a raging torrents of water through the usually safely navigable ford, sandbags versus shopping bags, vehicles laden down with loads which look like things which are not vehicles, and people frantically gesturing to stop.

Post reply on HN