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

121–130 of 146 posts

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

#121

Post by organsnyder »

joering2 wrote: post #75

Great to see such progress made by Waymo and Co. But as the Murphy's law says, if something might break, it will eventually break. With possibly millions of self driving cars and trillions of unique situations, its inevitable that someone will get hurt. So my question is: what kind of progress is being made to draft a legal framework for situation in which I rode my bike in bike lane and for whatever reason self-driv…

In your hypothetical situation, you'd probably deal with the owner's insurance company, just like with a human driver. The insurer will have already priced the policy lower based on underwriters' analysis that AI is safer than human drivers. In the case where the AI made an egregious error, I'm sure the insurer would be interested in pursuing legal action.

In my medium-sized metro area, I can recall at least a half-dozen deaths of cyclists this year caused by humans making similar errors. It doesn't take much for software to be better than that.

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

#122

Post by sangnoir »

nradov wrote: post #95
peoplewindow wrote: post #58

Earlier quoted context omitted.

I doubt Google will cut back any time soon, but as a former Google engineer, I find this an entertaining definition of business value ;) How much money does Google Photos make? Last time I used it, there were no ads in it. Isn't speech recognition nearly at human levels of comprehension, at least for US English and outside of highly technical jargon? Speech patterns change slowly, so would Android continue to make mo…

Speech recognition is still nowhere near human levels of comprehension. I have no speech impediments and speak US English with a neutral accent, yet Android speech recognition makes so many mistakes that it's usually faster to just type.

> 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.

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

#123

Post by andegre »

apk17 wrote: post #59
stevenwoo wrote: post #25

Earlier quoted context omitted.

I occassionally come across construction that is marked in an unusual way - at 65 miles an hour when the speed limit is 75. Also, when they are changing freeways and restriping them (leaving the original old faded lanes and before new lines go up) sometimes it's confusing to me and the traffic is flowing at 80 miles an hour, IIRC this was happening on 880 for a while about 10 years back. If the AI is confused by a ro…

They might want to look at https://www.google.de/maps/@48.4020174,9.9783793,159m/data=!... - the lower right part confuses drivers very much as well.

Wow, that would confuse me!

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

#124

Post by nradov »

sangnoir wrote: post #122
nradov wrote: post #95

Earlier quoted context omitted.

Speech recognition is still nowhere near human levels of comprehension. I have no speech impediments and speak US English with a neutral accent, yet Android speech recognition makes so many mistakes that it's usually faster to just type.

> 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.

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

#125

Post by danans »

peoplewindow wrote: post #58
dlubarov wrote: post #13

Earlier quoted context omitted.

Agreed. As a Google engineer, I think a lot of our AI efforts are "safe" (as in we'll keep investing in them for a long time) because they're already providing substantial business value. For example, - TTS and speech synthesis have lots of uses in Android (phones, Wear, Auto, etc.) - Object recognition is very useful for photo search - Face recognition is also useful for photo search (if you tag people in Google Pho…

I doubt Google will cut back any time soon, but as a former Google engineer, I find this an entertaining definition of business value ;) How much money does Google Photos make? Last time I used it, there were no ads in it. Isn't speech recognition nearly at human levels of comprehension, at least for US English and outside of highly technical jargon? Speech patterns change slowly, so would Android continue to make mo…

> Isn't speech recognition nearly at human levels of comprehension, at least for US English and outside of highly technical jargon?

Recognition of technical jargon is an easier problem than recognition common speech. Jargons are far more regular and have limited lexical and idiomatic variation, while common speech has a huge number of variations and is constantly innovating new usages. This is evidenced by the fact that the earliest speech recognition systems were only useful in highly restricted domains (medical transcription, phone support tree navigation).

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

#126

Post by notahacker »

Retric wrote: post #116
notahacker wrote: post #113

Earlier quoted context omitted.

I agree that self driving cars' response time can sometimes compensate for lack of general intelligence to anticipate a visible roadside activity developing into a hazard or non-routine situations in which another driver might cut into their lane. But lightning reflexes aren't going to eliminate situations in which buggy, late or nonexistent responses to things an AI hasn't been trained to deal with endanger other ro…

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 the football that flew past moments earlier, but a human capable of grasping that footballs are associated with pedestrians making rash decisions might have braked early and gently enough to not get bumped by the car behind. (If they didn't, they might find their late reaction blamed for the accident and possibly even get prosecuted for driving without due care and attention).

The UK requires every learner driver to sit an exam consisting of identifying CGI "developing hazards" where they're scored on ability to rapidly identify stuff that might happen before they're allowed to do the full driving test. I'm sure a key focus of the teams the article discusses is teaching AI similar cases like gently slowing in the event of a football-shaped object moving near the road (which is likely far from the most difficult or obscure novelty to teach an AI to handle) but the problem space of novelties humans handle by understanding what things might be and how/if they are likely to move isn't small or one there's good reason to believe plays to AI's strengths

(Meta: not sure why you're being downvoted, your contribution seems constructive and on topic to me)

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

#127

Post by jpindar »

jvolkman wrote: post #10
jacquesm wrote: post #9

Earlier quoted context omitted.

They don't have to simulate one, merely reliably detect one and then refuse to drive. One of the most important lessons drilled into new drivers is that the best way to avoid accidents is to know when not to drive, self driving cars should not be an exception to that rule. If you are going to attempt to drive in hurricane you are essentially committing suicide, your best cause of action is to find a safe place to wea…

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.

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

#128

Post by flamedoge »

jooke wrote: post #64
flamedoge wrote: post #57

> But Peng also presented the position of the traditional automakers. He said that they are trying to do something fundamentally different. Instead of aiming for the full autonomy moon shot, they are trying to add driver-assistance technologies, “make a little money,” and then step forward toward full autonomy. It’s not fair to compare Waymo, which has the resources and corporate freedom to put a $70,000 laser range…

Who's going to buy a car that might tell them they can't drive?

If it catches me dozing off during long drive, that's a perk in my view.

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

#129

Post by Animats »

dxbydt wrote: post #106
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…

> If you want to do it wrong, you start out with lane keeping...recognition of common objects...some machine learning....hope for the best Thank you! Can I ask why the self-driving car programs like the 3-semester Udacity one deliberately focus on going about it the wrong way ? Why spend 1 whole semester on lane detection using trivia like OpenCV, 1 whole semester on object classification using CNNs with Keras, spend…

All those subjects are useful. Some version of all those things will appear in the good systems. But it's the difference between building the flight control system for a toy drone and the flight control system for a Boeing 797. You need the "failure is not an option" attitude of aerospace. You need fault tree analysis. What if sensors disagree? What if a sensor is misaligned because someone bumped the vehicle while it was parked? You need enough of an organization to have people working through all those situations. You don't get to blame the user for your mistakes. A "minimum viable product" will kill.

Self-driving is something many people in software haven't seen much - software that isn't feature-oriented. As someone from Google pointed out, the output is just two numbers.

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

#130

Post by dsfyu404ed »

organsnyder wrote: post #120
dsfyu404ed wrote: post #93

Earlier quoted context omitted.

If you ignore low speed, low visibility parking lot stuff many people will go their whole lives without being in an accident that they are at least partially responsible for.

A "disengagement" isn't defined as an at-fault accident, though—it's a situation where the AI is unsure what to do, so it reverts to human control. I think the equivalent for a human driver would be a situation where they're unsure what to do, so they freeze up, or proceed extremely cautiously. I'm not prone to freezing up, but I've definitely encountered situations where I'm uncertain to the point that I proceed wit…

My point was that an at fault accident is an obvious case of a situation a human cannot handle and that it at least tells us something about the rate at which humans "disengage".

Human disengagement isn't easy to define. People don't just say "well this is too confusing, I'm just gonna turn control over to someone else". To define disengagement you need to define a performance minimum. An accident is about the lowest floor you can set to I used that. I would personally say that if you hit the brake on an on ramp while traveling slower than the highway traffic you're "disengaged." People and AI generally travel as fast as they can while maintaining a sufficient safety factor (e.g. people generally don't tailgate each other at triple digit speed and drive slower in snow). If someone's personal safety factor/comfort level requires several minutes of everyone else honking while they wait for a space fit for a semi-truck before pulling their Prius into traffic that may be acceptable to them but if an AI did that I think most people would say that's a failure even if the AI didn't turn control over to a human.

Post reply on HN