Waymo to customers: “Completely driverless Waymo cars are on the way” • Hacker News Board

Re: Waymo to customers: “Completely driverless Waymo cars are on the way”

#191

Post by Dylan16807 » Thu, Oct 10, 2019, 8:28 AM UTC

beerandt wrote: post #190

Dylan16807 wrote: post #183

Earlier quoted context omitted.

> 1) I'll rephrase: occasionally (and needlessly) stopping for a plastic bag (or any other false positive) should not be considered a massive improvement over current human abilities Okay, but I didn't say that either. I said it would be an improvement to the experience I have . I will trade an abrupt stop in front of a plastic bag for several hours of being able to do what I want while the car does the driving. > If…

>for several hours of being able to do what I want while the car One word: Chauffeur >The AI processing time to recognize things in a frame needs to match the framerate If you're talking about framerates at all, you're still missing the point. I'm talking about human ability to predict and anticipate. The fact that a human can brake before there's anything in the frame to even detect. We've (humans) got a head start,…

> One word: Chauffeur

But for people with normal incomes too!

> brake before there's anything in the frame to even detect

The scenario was slowing down a bit, but still needing to hit the brake in response to the child. So if the AI can brake faster it is extremely important.

Re: Waymo to customers: “Completely driverless Waymo cars are on the way”

#192

Post by Dylan16807 » Thu, Oct 10, 2019, 8:34 AM UTC

beerandt wrote: post #140

mgalgs wrote: post #135

Earlier quoted context omitted.

> That’s not image recognition. How do you do it? Telepathically? You're still just receiving visual input and processing it.

Human knowledge and intuition. Deductive reasoning. Knowing there is a location ahead without a crosswalk that kids constantly cross at. Or knowing that the car in front of you is going too fast, implying that they're unaware of the railroad tracks they're about to jump. You don't need telepathy to make useful predictions. And humans don't need to be accurate, they just need to be prepared to take a measured response…

> Knowing there is a location ahead without a crosswalk that kids constantly cross at.

That's not based on reasoning. Most people driving by probably don't even know. And it might be in the car's database.

> Or knowing that the car in front of you is going too fast, implying that they're unaware of the railroad tracks they're about to jump.

If I'm behind them, and I'm going much slower because I know the layout of the road... I'm not sure how knowing this matters?

Re: Waymo to customers: “Completely driverless Waymo cars are on the way”

#193

Post by Dylan16807 » Thu, Oct 10, 2019, 8:40 AM UTC

p1necone wrote: post #184

c0nducktr wrote: post #145

Earlier quoted context omitted.

Be a bullshit artist?

Not making promises you can't keep is being a bullshit artist?

Talking extensively about how wonderful your product will be without ever making any promises whatsoever is the work of a bullshit artist.

If you can't keep promises, and don't want to bullshit, then talk more about what the product already does.

Re: Waymo to customers: “Completely driverless Waymo cars are on the way”

#194

Post by misterman0 » Thu, Oct 10, 2019, 9:19 AM UTC

p1necone wrote: post #60

harry8 wrote: post #36

How to predict: 1) Say what but don't say when. 2) Say when but don't say what. Never, ever, ever bet tempted to say something specific will happen in a bounded time range. Flying cars are on the way! That disease you hate, a cure is coming! Next year will be a big year for other breakthroughs not mentioned here.

This is the golden rule of software dev too - promise features or promise a release date, but never both.

"promise features or promise a release date, but never both"

I agree and it's such a golden rule that no one follows or cares about even the slightest. Features and deadlines are to SE like velocities and positions are to QM.

This idea more thoroughly described here: https://news.ycombinator.com/item?id=21212087

Re: Waymo to customers: “Completely driverless Waymo cars are on the way”

#195

Post by pjc50 » Thu, Oct 10, 2019, 9:34 AM UTC

Dylan16807 wrote: post #111

2bitencryption wrote: post #61

Earlier quoted context omitted.

every now and then I do something while driving that, IMO, requires "human cognition". Not just image recognition, obstacle avoidance, rule following, etc. Like, I'm cruising along at 45mph and a plastic bag is blown in front of my car. Yes, I see an obstacle and need to make a decision about what to do - but I do this as a human who knows what a plastic bag is, and how it is far safer to drive right over it than sla…

Even if the car occasionally stops for plastic bags, the self-driving is still a massive improvement in the experience I have. For children you might preemptively slow down, but a properly designed system can brake a whole lot faster. Should be fine.

> can brake a whole lot faster

Up to a point - ordinary seatbelts even with smart pre-tensioners aren't great at this, and it may increase the rate of whiplash injuries for the passengers who aren't expecting braking.

Re: Waymo to customers: “Completely driverless Waymo cars are on the way”

#196

Post by losvedir » Thu, Oct 10, 2019, 11:16 AM UTC

revscat wrote: post #83

persistent wrote: post #7

Waymo is so far ahead in this game. Tesla has the biggest online cheerleading section and an "autopilot" that behaves as a low-budget adaptive cruise control with lane keeping that sometimes works, and a pretty big body count. Uber has a dead pedestrian and a major lawsuit. Cruise has a lot of people on staff but virtually nothing to show for it. Lyft exists. Waymo is in production revenue service.

> Waymo is in production revenue service. I’m going to ignore the fact that you’re flatly incorrect about the rest of your posts regarding Tesla, since others have more ably corrected you than I could, and simply ask what the hell this even means. A “production revenue service”? This sounds like meaningless babble. Waymo does not currently have anything available that you can buy, lease, subscribe to, or otherwise ex…

Nah man. They have a ride hailing app and seevice in Arizona. The email was to customers of this "production revenue service" that some rides soon will not have a safety driver in the car.

Re: Waymo to customers: “Completely driverless Waymo cars are on the way”

#197

Post by losvedir » Thu, Oct 10, 2019, 11:22 AM UTC

revscat wrote: post #100

persistent wrote: post #30

Earlier quoted context omitted.

Not sure there is enough evidence in support of those claims about Tesla. Tesla doesn't collect anywhere near enough data from its customer fleet to support the kind of massive training being claimed. In contrast Waymo and Cruise vehicles are regularly in the depot where each car offloads terabytes of data. Who has the bigger training set?

Tesla routinely gets data from the fleet and analyzes it. Here is Andre Karpathy speaking on this subject: https://youtu.be/Ucp0TTmvqOE 1:49:30. I’m on mobile otherwise I would link to the exact spot. Tesla has tens (hundreds?) of thousands of vehicles on the road right now that are continuously feeding training data back to Tesla. What evidence is there that this is insufficient, or even deficient? From what I can g…

Unless Teslas are uploading gigabytes every night over your home WiFi (do they?) they're limited by a 4G connection.

So you have a lot more Teslas sending a lot less data each. I'm not sure which company has the richer training set.

Re: Waymo to customers: “Completely driverless Waymo cars are on the way”

#198

Post by YeGoblynQueenne » Thu, Oct 10, 2019, 12:07 PM UTC

rubicon33 wrote: post #104

jonplackett wrote: post #71

Earlier quoted context omitted.

I think that sums it up really well. In those kind of edge cases like the basket ball, if an AI driver didn’t slow down and then killed a kid, we’d all be up in arms about It saying how obvious it was that a kid would run out, and we’d have stopped, and we’d be right. But at the same time that AI driver will have stopped in time when 10 other kids jumped out without any warning because it has much, much better reflex…

Nearly 1.25 million people die in road crashes each year. Think about that. Self driving cars aren't just a solution to nuisance of driving. They're a solution to one of the leading causes of DEATH and INJURY in the country. Sure there are going to be edge cases where we as humans will be able to point at the car and say "damnit, we could have prevented that"! But I sure hope when we do that, the car points right bac…

>> Self driving cars aren't just a solution to nuisance of driving. They're a solution to one of the leading causes of DEATH and INJURY in the country.

If reducing DEATH and INJURY are the motivation for self-driving cars, we can achieve a larger reduction with technology that is available right now: public transport. And there is even a no-technology solution: a drastic reduction of road traffic that passes through residential areas.

Putting more cars on the road might or might not reduce DEATH and INJURY, given that those cars are going to be self-driving. But _removing_ cars from roads will _certainly_ reduce DEATH and INJURY- and also reduce the environmental damage from the cars' internal combustion engines and the infrastructure needed for those cars to circulate.

Considering all of the above, I'm really not convinced that the self-driving car industry is really that interested in reducing DEATH and INJURY. I think their main motive is to increase PROFIT.

Re: Waymo to customers: “Completely driverless Waymo cars are on the way”

#199

Post by YeGoblynQueenne » Thu, Oct 10, 2019, 12:15 PM UTC

jonplackett wrote: post #42

I have very mixed feelings about autonomous cars. One side of me is so excited about the prospect and finds it all amazing and futuristic. I love reading about how Tesla and Waymo approach it differently. The other side of me is absolutely not ready to trust my life to a machine - even though I know I do that all the time in other ways in modern life. I think it will be people's emotional evaluation that will matter…

>> Driverless cara will probably be safer statistically but they’ll simultaneously make errors a human wouldn’t, so some people are going to die in what will seem like really dumb ways.

Statistically it would take tens or hundreds of years before we can know whether self-driving cars are safer than human-driven cars:

  Key Findings

  ∙ Autonomous vehicles would have to be driven hundreds of millions of miles
    and sometimes hundreds of billions of miles to demonstrate their reliability
    in terms of fatalities and injuries.

  ∙ Under even aggressive testing assumptions, existing fleets would take tens
    and sometimes hundreds of years to drive these miles — an impossible
    proposition if the aim is to demonstrate their performance prior to
    releasing them on the roads for consumer use.

  ∙ Therefore, at least for fatalities and injuries, test-driving alone cannot
    provide sufficient evidence for demonstrating autonomous vehicle safety.

From:

Driving to Safety

How Many Miles of Driving Would It Take to Demonstrate Autonomous Vehicle Reliability?

https://www.rand.org/pubs/research_reports/RR1478.html

Re: Waymo to customers: “Completely driverless Waymo cars are on the way”

#200

Post by eloff » Thu, Oct 10, 2019, 1:16 PM UTC

jdashg wrote: post #187

eloff wrote: post #51

Earlier quoted context omitted.

You'll never run out of edge cases if you have to explicitly program the logic. At some point you have to let the machine learning decide it. Maybe there is room for a hybrid approach, but I think anything relying on programmers having thought of every situation ahead of time is doomed to fail.

I believe relying on machine learning is doomed to fail because I don't think it can feasibly offer the robust safety guarantees required. I honestly don't expect go-anywhere L5 in the next decade at least, and I definitely expect an engineered approach to hit L4 first.

You're wrong about the requirements. It doesn't have to be perfect, just better than us, and that's easily doable for computers that don't get tired, distracted, or drunk.

I'm not sure how long it will take, but I'm sure we'll get there.