Hacker News Reader

Comment by codampa01314 | Hacker News Reader

> In San Francisco, those problems can mean self-driving cars blocking traffic, transit and emergency responders, as well as erratic behavior resulting in close calls with cyclists, pedestrians or other vehicles.

I have mixed feelings about how to test self-driving vehicles (mostly stemming from ignorance). But at some point don't you have to get this stuff out in the wild to see how it behaves or else we are resigned to not making progress on this (or very very slow progress)? And considering that "no one was hurt" and "driverless taxis have never killed or seriously injured anyone in the millions of miles they’ve traveled" are were now there?

View on HN

> at some point don't you have to get this stuff out in the wild to see how it behaves or else we are resigned to not making progress on this

It's not my problem whether Waymo makes progress on their technology, but they make it my problem when they hit me with their car.

I do have sympathy for wanting to make the progress: we might make better pacemakers faster by testing out 20 designs on 20 people. But nobody wants to be the one that gets the one that doesn't work. And in this case the victims aren't even involved. They didn't opt-in to beta testing the road full of half-baked robots, they're just trying to get to work.

In the interest of having a discussion, let's assume that AVs are a meaningful goal to work towards for whatever reason. But given that, how do we get vehicles from the drawing board to actually being viable products without on-road testing? I've never seen a complex product that went from non-existent to perfect in the first deployment, so it doesn't seem realistic to expect that here.

Instead, they should be developed iteratively, with design prototypes that proceed from closed-course testing to supervised public testing to closed course autonomous testing, to on-road autonomy over the course of many years. This is what Waymo did. There's a reasonable argument to be made that they did this too quickly, but I can't reconcile that with your argument that they shouldn't have done it at all.

In an ideal world, there'd also be effective government oversight and public safety monitoring at every stage of the above process. Regulators haven't stepped up to do this, though AV companies have done quite a bit to stymie the oversight process as well.

Sadly, I come only with problems and not solutions. I take it as axiomatic that beta-testing with peoples' lives that didn't agree to do so is unacceptable. That closes off a lot of the solution space that you're proposing. That sucks and you're free to disagree but again I take it ethically unassailable.

Teleportation would also be a societal game changer but if the only way there is to beta test it on unwilling participants I'd also believe that, well, we just don't get teleportation then.

It's up to Waymo to figure out how to get there, not to me. I do not take it as axiomatic that just because it'd be useful that the ends justify the means. And it certainly isn't up to Waymo whether you or I can be sacrificed.

Should student drivers not be allowed on the roads either?

I don't take these to be analogous. Even the worst human driver can recognise that they don't recognise a situation.

And Waymo and Cruise are only rolling out gradually into settings they have reason to think they can handle. And unlike for humans, every time they learn a new lesson, the improvement spreads to their whole fleet. Humans lack that ability -- in a sense, we're the ones who are untrainable.

(If you want to point at Uber before they stopped, I agree they were irresponsible.)

They've been testing this before putting it on public roads for what, 12 years?

How much longer do you think they should keep testing it off of public roads beyond when they think they're ready?

> It's not my problem whether Waymo makes progress on their technology, but they make it my problem when they hit me with their car.

Have they hit anyone with their car? If they have, have they done so at a rate higher than a human driver?

What if they make it less likely for you to be hit by a car? Doesn’t that give you a benefit rather than a problem?

Uber’s driverless vehicle killed someone jaywalking at night in another state. So you can’t say they’ve never killed anyone. Plus Teslas have killed their own occupants plenty of times.

I'm debating about renewing my truck license for this reason. If it's taking this long for automobiles to be approved and accepted, it'll be 30 years before trucking is automated.

The two decades during which human oversight of automated systems will be mandatory would be long enough for me to finish off my career getting paid to drive while I sit in a cab writing code, periodically checking over the status of my lead truck and the two or three slaved trucks following me.

I think there will be a strong demand for truck drivers long after most people would expect.

We’re far from self driving vehicles still, and trucks will be the last to be automated, and once they are, I expect they’ll require human supervision just like you predict.

I'm debating about renewing my truck license for this reason. If it's taking this long for automobiles to be approved and accepted, it'll be 30 years before trucking is automated.

Trucks and cars are different. They've been running automated big rigs between Dallas, Houston, San Antonio, and El Paso for a few years now.

I hadn't heard that, but a cursory Google search indicates that Kodiak is building one of these systems.

https://kodiak.ai/

Given the significant weight and danger of death for other drivers, it'll be years until legislators allow their "safety drivers" to be eliminated from the equation. This makes the AI system more akin to enhanced cruise control than robotic trucking.

It's been in the newspapers down there quite a bit. Mostly Houston and San Antonio.

Here's one article about Volvo:

"Companies such as IKEA, UPS and FedEx have begun using autonomous trucks to make long haul trips across Texas, most often along I-45 between Dallas and Houston. In 2017, the U.S. Department of Transportation designated Texas as one of that nation's 10 proving grounds for autonomous vehicle testing. And Texas lawmakers have encouraged autonomous-vehicle development by making sure traffic laws do not encumber the companies." https://www.houstonchronicle.com/business/retail/article/vol...

Not disagreeing, but you might find it interesting to learn about the history of the term "jaywalking"

This is a good point. The statement I made about not killing anyone (in quotes) is from the article, which is only a subset of the overall domain we're talking about.

Reply on news.ycombinator.com