Live data from Hacker News

Self-driving Uber car kills Arizona woman crossing street

reuters.com

611–620 of 1001 posts

Re: Self-driving Uber car kills Arizona woman crossing street

#611

> You can't really get away from this without putting these cars on rails. We should just really build rails. Self driving cars are dumping billions of dollars onto a problem better solved by rails. Several cities already have self driving trains including London, Singapore and Kuala Lumpar, and they transport millions of people every day without incident: http://penguindreams.org/blog/self-driving-cars-will-not-sol.…

I don’t oppose public transport at all and the US has a lot to improve in this area. But e.g. East Asia and Europe have plenty of rails, and people still want to drive their own cars, and there are still an unbelievable number of accidents. We should require that every traffic participant gets a sensor, and connect all in a (local) network, instead of only relying on machine learning and its vulnerability to the unpr…

Germany here, we don't actually have "plenty" of rails for public transport, we love our Autobahn (no general speed limit!) and our cars, yet our vehicle fatality rates are among the lowest in the world (half that of the US).

The reason that fatalities are 10x higher in Asia is that people drive (and walk!) like lunatics and vehicle fatalities are just shrugged off.

Just be more orderly is basically what I am saying. Also eat more sauerkraut - it's healthy!

Re: Self-driving Uber car kills Arizona woman crossing street

#612

> You can't really get away from this without putting these cars on rails. We should just really build rails. Self driving cars are dumping billions of dollars onto a problem better solved by rails. Several cities already have self driving trains including London, Singapore and Kuala Lumpar, and they transport millions of people every day without incident: http://penguindreams.org/blog/self-driving-cars-will-not-sol.…

You're right. There's a lot of research and experience that suggests a well-developed rail network would have all sorts of benefits. We could take it a few big steps further and point to other problems like suburban sprawl in America, dying small towns, etc.

But there are a couple of really big, really massive problems. We have trillions in existing infrastructure and trillions more in existing property that's been developed over the course of decades. A lot of that development has come with unintended consequences. Just look at Houston after Hurricane Harvey and how their development and zoning practices took the insane rains Harvey deposited and magnified their negative consequences.[0] Decades of short-sighted decision-making took a hurricane and made it worse. Some changes will likely be made, but they won't solve nor likely keep up with a problem that's only going to get worse. That'll happen because the alternative, demolishing huge portions of the sprawling concrete city, isn't economically viable. The city can't be moved.

It's baggage. And there's so much more. Tens of trillions of dollars worth. Then there are existing roads, bridges, and other infrastructure connecting all of it. Designed for cars, and much of it legacy infrastructure that doesn't take into account modern infrastructure design principles. Add cities and towns, some of which are (charitably) in less-than-optimal positions. Towns, strip malls, and other structures that wouldn't exist if it weren't for suburban sprawl. Baggage. And then there's the people. People with ideas on what their lives ought to be like and where they want to live. There's a town, Centralia, Pennsylvania, that is literally on top of a massive underground fire that's been burning for over half a century. There was a massive relocation effort with federal and state funds. The town has no services, its zip code was discontinued in 2002, and it's still burning with all of the potential health risks that involves. There are still five residents there today, who have fought every attempt to relocate them. They will continue to live there until they die, at which point their homes will finally be taken and demolished through eminent domain.[1] If we can't bulldoze a town like that, we sure as hell can't take a bulldozer to normal ones.

Baggage. What it means is that we can't simply replace everything. It can only be done piecemeal, which means new infrastructure must take into account the need to accommodate everything that was developed under the old. High speed rail can be built out in the United States, and I genuinely hope that it is. But it literally can't replace the car in its entirety.

Self driving car research is--compared to the cost of major infrastructure replacement, let alone an alternative that simply isn't feasible--practically a rounding error. It doesn't hurt us, because the option was never a binary choice: self-driving cars OR rail. Self-driving cars give us the ability to deal within existing legacy constraints as they're slowly adapted, modified, and/or replaced. They also offer some long-term possibilities that will allow us to make radical changes in urban design. And yes, those changes are ones can accommodate rail, pedestrian-friendly city design, and other options that can drastically improve quality of life.

0. https://www.npr.org/2017/11/09/563016223/exploring-why-hurri...

1. https://finance.yahoo.com/news/pa-residents-living-above-min...

Re: Self-driving Uber car kills Arizona woman crossing street

#613
post #596

There are roughly 3e8 cars in the US, they kill 3e4 people a year => each car kills 1e-4 person a year. If we have say 100 AV right now in US, each car kills 1e-2 person a year. So that's 100 times more dangerous than humans...

That's not a good comparison because self driving cars are on the road a lot more often than regular cars.

Re: Self-driving Uber car kills Arizona woman crossing street

#614
post #582

> You can't really get away from this without putting these cars on rails. We should just really build rails. Self driving cars are dumping billions of dollars onto a problem better solved by rails. Several cities already have self driving trains including London, Singapore and Kuala Lumpar, and they transport millions of people every day without incident: http://penguindreams.org/blog/self-driving-cars-will-not-sol.…

I couldn’t put my finger on it at first, but I then realized why your comment doesn’t make any sense - you’re effectively arguing “if I could master-plan society and the economy for a while, this is how I’d decree it to be.” Maybe some people don’t want to live in dense urban areas, maybe they don’t want to ride on a rail, maybe they want privacy. But looking out the window, I can say American society has soundly rej…

How can you reject something you never had or used? Almost nobody alive in the US today has ever engaged with general commuter rail. Trains as a common mode of transit died off with the rise of cars and planes post WW1 and had them near-eradicated post WW2.

You can't say people reject something they don't even know is a possibility.

Re: Self-driving Uber car kills Arizona woman crossing street

#615

> You can't really get away from this without putting these cars on rails. We should just really build rails. Self driving cars are dumping billions of dollars onto a problem better solved by rails. Several cities already have self driving trains including London, Singapore and Kuala Lumpar, and they transport millions of people every day without incident: http://penguindreams.org/blog/self-driving-cars-will-not-sol.…

in my area, our rail proposal was 1 billion for 9 miles of trail. Good luck putting everything on rails.

Ridesharing algorithms will 100% solve the problem of underutilised mass transit. Self driving isnt necessary.

Imagine shuttles that hold 10 people using uber pool algorithms and funded by the city.

People will be able to make their request and a nearby shuttle will pick all the people up in that area. The shuttles could do ad hoc transfers with other shuttles so people wouldnt even have to be going to the same area.

The algorithm would minimize the average ride time for everyone.

The #1 issue with mass transit is lack of convenience. Rideshare algorithms will eliminate that lack of convenience.

Re: Self-driving Uber car kills Arizona woman crossing street

#616

This is what's going to happen. If you've ever seen a machine learning algorithm in action, this isn't surprising at all. Basically, they'll behave as expected some well known percentage of the time. But when they don't, the result will not be just a slight deviation from the normal algorithm, but a very unexpected one. So we will have overall a much smaller number of deaths caused by self driving cars, but ones that…

May be for now, autonomous driving should be limited to freeways and roads that dont have pedestrian crossings/pavements.

Re: Self-driving Uber car kills Arizona woman crossing street

#617
post #582

> You can't really get away from this without putting these cars on rails. We should just really build rails. Self driving cars are dumping billions of dollars onto a problem better solved by rails. Several cities already have self driving trains including London, Singapore and Kuala Lumpar, and they transport millions of people every day without incident: http://penguindreams.org/blog/self-driving-cars-will-not-sol.…

I couldn’t put my finger on it at first, but I then realized why your comment doesn’t make any sense - you’re effectively arguing “if I could master-plan society and the economy for a while, this is how I’d decree it to be.” Maybe some people don’t want to live in dense urban areas, maybe they don’t want to ride on a rail, maybe they want privacy. But looking out the window, I can say American society has soundly rej…

> maybe they want privacy

Every sci-fi movie of the last 50 years has solved that problem: you don't use trains; you put the cars on the rails (by grouping them together into roller-coaster-looking setups.)

> maybe they don’t want to ride on a rail

Same here: cars on rails can, in most conceptions, signal to the rail system to switch them onto an offramp that dumps them back onto a road.

Basically, here, "rails" are serving as safer replacements for highways. Every road that isn't a highway would still be a road. You're just enabling people to, effectively, "have their car take the train", rather than having to leave their car at the train's parking lot.

Re: Self-driving Uber car kills Arizona woman crossing street

#618
post #582

> You can't really get away from this without putting these cars on rails. We should just really build rails. Self driving cars are dumping billions of dollars onto a problem better solved by rails. Several cities already have self driving trains including London, Singapore and Kuala Lumpar, and they transport millions of people every day without incident: http://penguindreams.org/blog/self-driving-cars-will-not-sol.…

I couldn’t put my finger on it at first, but I then realized why your comment doesn’t make any sense - you’re effectively arguing “if I could master-plan society and the economy for a while, this is how I’d decree it to be.” Maybe some people don’t want to live in dense urban areas, maybe they don’t want to ride on a rail, maybe they want privacy. But looking out the window, I can say American society has soundly rej…

Maybe American developers have rejected the rail, but I don't think American passengers have.

Re: Self-driving Uber car kills Arizona woman crossing street

#619

This is what's going to happen. If you've ever seen a machine learning algorithm in action, this isn't surprising at all. Basically, they'll behave as expected some well known percentage of the time. But when they don't, the result will not be just a slight deviation from the normal algorithm, but a very unexpected one. So we will have overall a much smaller number of deaths caused by self driving cars, but ones that…

Very little of what goes into a current-generation self-driving car is based on machine learning [1]. The reason is exactly your point -- algorithmic approaches to self-driving are much safer and more predictable than a machine learning algorithms. Instead, LIDAR should exactly identify potential obstacles to the self-driving car on the road. The extent to which machine learning is used is to classify whether each ob…

Actually, because there's a severe shortage of LIDAR sensors (much like video cards & crypto currencies, self driving efforts have outstripped supply by a long shot), machine learning is being used quite broadly in concert with cameras to provide the model of the road ahead of the vehicle.

Re: Self-driving Uber car kills Arizona woman crossing street

#620
post #602

The title here should be amended to “...hits, kills jaywalking pedestrian...”. Let’s get the facts straight. This person was walking outside the crosswalk. The fact that a computer was behind the wheel is irrelevant. A human broke a law designed to prevent this exact scenario, and the result was both predictable and tragic. It’s terrible, but ultimately the person’s fault.

The law is that pedestrians always have right of way. Jaywalking is a fine, but not a jail sentence. Not coming to a full stop for jaywalking pedestrians, however, is how I failed my first driver's test.

The law is that pedestrians always have right of way

That actually depends on the state. “Arizona law requires pedestrians to cross the street at corners and/or designated crosswalks...[if] it seems unlikely that a driver should have avoided---as in a case where the pedestrian darts right out into traffic, then it may be that the pedestrian will be deemed 100% at fault for his/her injuries.” [1]

I’m sure that Uber will be sued over this, and they’ll probably settle quickly and quietly to avoid mass hysteria resulting from misleading headlines such as this one. But that doesn’t mean the jaywalking pedestrian isn’t at fault. Walk in traffic, die in traffic. It’s a pretty straightforward physics problem.

[1] http://www.zacharlawblog.com/2015/11/whos-at-fault-in-an-acc...

Post reply on HN