Hacker News new | past | comments | ask | show | jobs | submit login
Google lobbies Nevada to allow driverless cars (nytimes.com)
133 points by nkassis on May 11, 2011 | hide | past | web | favorite | 101 comments

I think it's sign of our technologically advanced times that the commentary here is so level-headed.

I, for one, can't help having the mind boggled. Cars that drive themselves coming finally coming out of the domain of Knight Rider, and onto actual roads?! That's just frickin awesome! (or maybe I've just been watching this Louis CK clip too often - http://www.youtube.com/watch?v=8r1CZTLk-Gk)

One interesting aspect about driverless cars is that they can be used for traffic optimization.


tl;dr driving smoothly and leaving traffic gaps ahead can prevent traffic jams or slowdowns

Having just a small percentage of automated cars with traffic optimizing behavior may actually increase road capacity.

> tl;dr driving smoothly and leaving traffic gaps ahead can prevent traffic jams or slowdowns

An other traffic optimization (which doubles down as a gas optimization) is "car trains": cars following one another very closely (much more closely than human drivers would handle)

I think it could help solve congestion as a problem too. Maybe cars could book their routes, and the road brokers will maximise for #cars/destination instead of allowing people to keep piling in even after the useful capacity is exceeded.

That would probably be the hardest problem to solve, and it would require global cooperation, but yeah it would also be a nice end-goal.

Cooperation would already go a long way to solving congestion with just local cooperation though:

* higher road density (cars can pretty literally travel bumper-to-bumper)

* better handling of congestion (car train can smooth local slowdowns over all its component cars)

* much, much improved merging and splitting of roads (including local flow analysis e.g. a small road merging into a bigger interstate with a 3:1 traffic ratio could be managed into a "perfect" 3:1 merge frequency)

* simplified urban planning (if only cooperative automated cars are on the road, you can get rid of red lights)

You can only get rid of red lights if you don't care about pedestrians. I hope automated cars don't lead us to exacerbate the car-centric city problem we've built for ourselves.

That is a good point.

This won't be high on the feature list, as it's still really dangerous.

> This won't be high on the feature list

You could hardly be any less correct. This is by far the most major and interesting feature enabled by auto-driven and cooperative cars: http://www.sartre-project.eu/en/Sidor/default.aspx

> as it's still really dangerous.

Not really.

If you're adding laser rangefinders and a trunk full of computer equipment, you might as well weld on a few mechanical couplers, too. They work well enough for trains, anyway.

IMHO, the long term implications of driverless cars are far bigger than just traffic optimization. The article mentions ZipCar, and services like a robotic ZipCar or robotic taxis are what will really revolutionize transportation. Right now taxis are only convenient in certain high density areas (city centers, airports, convention centers, etc.). Try to call one from the suburbs and the not only is the wait long, but the fee is rather high also. But with a robo-taxi, it seems like it might be cost effective for the taxi company to leave some idle taxi's in shopping mall parking lots, etc. around the city, thereby making it possible to respond to a passenger request very quickly and (hopefully) inexpensively.

It might also increase the capacity of human productivity. I've always been amazed at the amount of collective concentration that keeps everyone in their lane and not hitting one another. Just imagine if you could channel that concentration into productive endeavors, such as answering emails, looking for a better job, or learning a new skill.

..., Watching reruns of Lost, playing Solitaire, napping, reading HN, ...

Sorry, I should have said "channel some of that concentration into productive endeavors"

As a someone who commutes by train every day, I assure you that this is much, much more likely than any productive use of the time.

Being able to turn off your brain for an hour a day definitely boosts productivity.

And emotional stability. When I've had a rough day at work, an hour of downtime on the train really improves the face I show at home.

I am always impressed by this when I read it--not because of the cars but the motivation. Here's a company that is so desperate to increase its ad revenue that it's willing to build robot cars so that it has that many more eyeballs to serve up advertisements to.

Here's a company that is smart enough to know that they could have a major part of a market though could explode. Google don't just have to sell ads. They have loads of very smart people. They could wind up being the major provide for driverless cars.

Sebastian Thrun works for Google now. He's one of the world's foremost experts in the field.

"They could wind up being the major provide for driverless cars."

I'd say they could be the major provider for driverless car technology. I don't think they should get into actual car manufacturing.

Bingo, this is going to be a big field in the future. AI is outside the core competency of car manufacturers; they'll have to buy that tech from some one. Google's ad revenue could look puny by comparison some day.

This sounds like the history of a company out of a Stephenson or Gibson book. I want to believe...

Indeed. This is what I meant.

This could also be part of Google looking for revenue sources beyond advertising. As other commenter have pointed out, if Google can make this work and smooth out the very non-trivial legislative and liability issues, there's an absolute fortune to be made licensing this technology to car companies. Think about it. If the technology works, and it's legal in your state, do you know anyone who wouldn't want a self-driving car?! It would revolutionize commuting. You could nap, read, watch TV on your phone or tablet, etc. instead of staring in boredom at the bumper ahead of you.

Google looking for revenue sources beyond advertising.

Naw. They are hoping that instead of spending your time behind the wheel, you will be spending it on the internet watching some Google ads.


They probably build robot cars because they are an artificial intelligence company.

Yes. They probably can make quite some money by patenting that tech and licensing the components to car makers.

Why be so cynical? Here is a company that has found such a massive golden egg that at least for now it can hire really really smart people to work on cool projects that may end up changing the world.

I think they just like playing with stuff that is this cool. It's great that it aligns with their business model, but I don't think the thought process went "hey, let's build a driverless car so people can see look at ads more!"

It's also a good recruitment tool having a couple of world-class people working on stuff that interests a lot of computer science types.

No, they are building robot cars because

1) They can sell them (the market is enormous)

2) It's fun and interesting

3) Great for the overall company image (as innovators)

Google's research into driverless cars really boosts their perception of "info hegemon having their hands in everything" and either "rock star company trying to save the world with bleeding edge technology" or "Orwellian nightmare come to life doling out Trojan horses so the masses fail to see the inevitable robot holocaust", depending on your personal bias.

That said, why is Google the only high-profile company looking into this? I've heard about the DARPA Grand Challenge, but that always seemed to be about hobbyists and university teams. I see there are quite a few groups interested in the concept, but it seems like only GM is the other company who bothered looking into it: http://en.wikipedia.org/wiki/Driverless_car#Key_players I suppose car companies are set in their ways and don't want to create a future that takes drivers out of the equation (though it's not as if people won't be flocking to buy driverless cars anyways), but why not any other tech companies?

The head of the Stanford DARPA Grand Challenge (Sebastian Thrun) left his tenured professorship at Stanford to work at Google.

The hobbyist/university teams are an excellent staging ground to get started; but any other company could have put in the investment that Google did. I would have loved to have seen a CMU/Red Whittaker/GM collaboration. It would have made for a poetic little "revive the rust belt with the robot car industry" story too. Given the response to the Detroit Superbowl Chrysler ad, looks like the public love that sort of story...

I anticipate a lot of luddite rage and probably a good number of accidents (and maybe even a few deaths), but we're not going to get to The Future by just wishing it so.

Human drivers kill ~260,000 children a year. The bar for robots is low.

Yes, but robots only need to kill one to be demonised. I think this is because the human theoretically can regret it and be punished, whereas there can never be a feeling of justice with the robot. The manufacturers take on a lot of liability (GOOGLE'S CHILD-KILLING DEATH MACHINES!)

Another thing is that any accident that an autonomously driven car gets into will be blamed on the car automatically, before any information is available.

In other words, it probably only takes an accident, not even culpability, to get demonized.

I think that autonomously driven cars will have a lot of cameras and a black box like airplanes do, so that all the data before the accident will be available and if somebody else is guilty of accident - videos will be published by car manufacturer.

The videos are hopefully going to carry some weight, but I find your consolation in the available data somewhat optimistic; at best, I think they'll just cushion the backlash.

And yet individual regret / manslaughter charges don't seem to actually change road death statistics. I'd so love for this to happen; I kinda also doubt it will though anytime soon.

That sounds like an awfully high number. Is there a source for it?

It is not very high if you consider that people drive cars everywhere, not only in the US.

The fatality rate in the US is around 30k-40k every year[1].

[1] - http://www-fars.nhtsa.dot.gov/People/PeopleAllVictims.aspx

I wonder if making robots drive would bring the traffic death rate down to the one of other industrialized countries.


Why is the rate so much higher in the US?

First thought would be to graph it against miles driven per person per year.

What makes you think machines would be even deadlier drivers than humans?

I read that comment as saying that people would get upset about the driverless cars and there would be bad PR for every accident, not that the driverless cars would actually be less safe. They won't be perfect, and every single failure will place the blame on the manufacturer, similar to the difference between public perception of airplane safety vs. car safety. Most people feel safer in cars, because they're in control. But people are safer in airplanes.

Honestly, I predict that people will hate the way driverless cars drive, as well. For many reasons, including liability, such cars will have to drive in a nicer (more defensive) way than most people do.

A driverless car won't tailgate and so it let people in, even if they're "cheating" in heavy traffic. Some people will absolutely hate that, even if it helps traffic flow better.

Hopefully, people will be too busy surfing the web to notice that their cars are politely improving traffic flow.

> Hopefully, people will be too busy surfing the web to notice that their cars are politely improving traffic flow.

I actually did that while traveling to work this very morning :) No need to dive into the future for that. It's called public transportation.

I drove to work today because my wife ankle was sore. I have a book I've been trying to finish and the 30 minute drive sorta annoyed me. I could have been so much more productive during that time.

I also drive from Canada to Florida multiple times a year. 24 hours wasted but it's better than flying, more freedom to stop and relax, better food and less stress overall and more space to carry stuff. With these cars I'd be in heaven (not literally).

The people complaining will be those who drive their own cars and are paying attention to how the computers drive.

They will complain that the robots are driving so nicely and defensively? That seems unlikely.

No, there will be complaints about how slow they drive because, without a doubt, there will be legislation requiring these cars not exceed the maximum posted speed limit.

True, but it could be mitigated by making the computer automatically take a picture of the plate in front.

The car computer saying "yeah, that just cost him a 50 dollar fine" would be absolutely loved by its users.

And it would force the other drivers to drive by the rules.

To be fair with the airplane example, if something goes wrong in an airplane there is the waiting period as the plane plummets to the ground, whereas a car accident would be over very quickly (unless you started rolling down the side of a tall mountain or something, I guess).

Here's the thing: driverless cars right now presumably are less safe. Last I heard they could follow lines on the road and dodge witches' hats, but they can't deal with unexpected situations.

Human drivers may be crap (though a human driver not actively being an idiot can generally travel many millions of miles between serious accidents) but I think it'll take a lot of R&D before robots can match 'em.

edit: Really? I got modded down for this? Go talk to one of the engineers on the Google driverless car program about how long he thinks it'll be before driverless cars are safer than the median human driver on a busy suburban street.

The thing you're not addressing is that there is a wide range of conditions and activities which all fall under the title of "driving". Would I trust a driverless car with current technology to navigate a busy parking lot? Not a chance. Would I trust it with the 40 mile commute I had, 38 miles of which were 6 lane highway? Probably. Would I trust it to make that same commute at times of low traffic when I'm likely tired myself? Absolutely. Not all driving needs a human's full attention, and the situations where a human doesn't need to pay full attention can be handled better by a computer which is immune to many of the distractions and weaknesses of the human mind. It's all situational and difficult to figure out, but the proper response is to keep plugging away at the parts of the problem you can gain some traction on and take what you can get.

There are cases like the busy parking lot where human control will likely be necessary well into the future, there are cases which can be addressed today, and there are cases which were addressed successfully over 20 years ago. Look up Ernst Dickmanns - he was an early pioneer in this field, and had an autonomous vehicle driving on European highways in traffic in the 80's. It's really easy to say that this is an impossible problem to tackle, but the person who says something can't be done should make damn sure he's not getting in the way of the person doing it.

I did my M.S. on this topic, so although I would not claim to be an expert in the field at Google's level, I do believe strongly in the technology and have given it more than cursory thought. The larger problems are more societal than technological, and as such I really hope that Google makes this happen because they have the societal clout to get people on board with the idea that a computer can, for most of the miles you drive, do a better job of getting you from A to B than you can.

Would I trust a driverless car with current technology to navigate a busy parking lot? Not a chance. Would I trust it with the 40 mile commute I had, 38 miles of which were 6 lane highway? Probably. Would I trust it to make that same commute at times of low traffic when I'm likely tired myself? Absolutely. Not all driving needs a human's full attention, and the situations where a human doesn't need to pay full attention can be handled better by a computer which is immune to many of the distractions and weaknesses of the human mind. It's all situational and difficult to figure out, but the proper response is to keep plugging away at the parts of the problem you can gain some traction on and take what you can get.

Sure, I think we're in complete agreement here. But the idea raised in the article that we're going to see driverless cabs on the Las Vegas Strip? I really don't think so. The day when we can trust a completely driverless car to wander round on its own without hitting anything is a helluva long way off.

In the meantime my main concern is that the hybrid human/computer driver might wind up with the worst features of both. It'd be much like an airliner, which can pretty much fly itself from origin to destination 99% of the time, but needs a carefully trained pilot for the occasions when something goes wrong... except that when an unexpected situation crops up on the road it requires a split-second decision, not the several minutes which airline pilots usually get.

Go ask a biologist how long it'll be until humans evolve 360-degree vision.

Like airplanes, they would be safer than driving. But, each incident will receive so much more coverage that the statistical safety will be largely ignored by the terrified public.

Airplanes still have pilots. I guess busses are safer than cars too.

i don't think its about them being deadlier. i mean, just look at the toyota "runaway" car issue.

all it takes is 1 car out of millions to cast doubt on the whole concept, or one person to do something stupid and blame it on the car to cause a PR firestorm.

I wonder how they're getting over any stigma culture shock. In their own products, they're careful to reduce any "creep factor" by not using our information in ways that provide too much immediate feedback, like ads that are too highly targeted. Unlike ads, a "robot car" is a significant jump. In the public's eye, I think Google's driverless cars last year seemed like a sudden leap.

There are obviously safety measures that will be in place, but I want to know how robust they're going to be and how much PR they're gonna put into this.

I hope some smart people are trying to figure out how our economy is going to work when transportation is outsourced to robots. Obviously it'd be great but the idea that everyone needs a job is going to start unravelling.

Its been that way for decades. Carpet used to be the most luxurious thing in the world; as soon at it got cheap, the old, undesirable hardwood became the preferred sign of wealth.

If robots start taking over a lot of the work, people will start wearing coats made out of solved captchas or something.

I have thought of driverless buses for a while now. These may be even better than driverless cars because buses always follow a defined route.

I would be more inclined to try out automated light rail/subways first, seeing as they don't need to navigate traffic.

The Copenhagen Metro is driverless (and there are probably many others around the world). Works like a charm.

The [London Docklands Light Railway](http://en.wikipedia.org/wiki/Docklands_Light_Railway) is also driver-less. However travelling everyday on it, the journey is not smooth. Especially when it starts or stops, the acceleration and deceleration is very jerky. And also the train lurches forwards quite a few times.

Not what one would expect on any train especially an automated train.

Vancouver's SkyTrain ( http://en.wikipedia.org/wiki/SkyTrain_(Vancouver) ) is driverless and very smooth. I have a feeling the unpleasantness of the ride of the LDLR has more to do with characteristics of the track or technology, rather than the driver's actions.

After riding this train I started to wonder why driverless trains are not more prevalent. It seems so obvious now.

You can't tell the transit unions, "you all have 1 year of employment left before we replace you with computers."

I don't know what it's like on the LDLR you mentioned, but I regularly ride the (very) human dependent 'L' in Chicago, and that's not a smooth ride either. So I'm not sure how much it has to do with the computer drivers.

Could this not be the same thing you sometimes get on the human-driven tubes - for some reason (bizarre safety?) apparently if people lean on the doors they brake suddenly (have heard drivers say that over PA, googled for some kinda explanation but couldn't see anything except speculation).

You'd think it'd be cheaper and more reliable to make the doors sturdier.


Sounds like Vegas would be a good starting point, no one remembers what happens there (zing). In all seriousness, Google is going to drop a ton of money into Nevada to make this happen and drive more people to Vegas again. I call this a stimulation package, hope it passes.

I don't think Vegas would be the starting point. Nevada is in the midst of massive unemployment numbers and a massive budget crises. There are a lot of people employed in Vegas as drivers of cabs, limos and other private services. Launching driverless car services in Vegas would likely meet with massive public resistance. I could see it as another tourist attraction for the strip I suppose but I wouldn't want to test autonomous vehicles there given the crappy traffic and idiotic/distracted drivers there.

I would probably start testing in the Carson City, Reno, Sparks area. You can get a flight from the bay to Reno that takes 45 minutes. If you need to take cars back and forth to Mountain View it is only a 3-4 hour drive. The capital is Carson City so if you need access to the legislature you have it. There's also very diverse geography and lots of public roads available for testing. If you need populated areas or unpopulated areas everything is here all while being a lot closer to Google than Las Vegas is.

California's budget deficit is larger, but a large sum of money from Google can probably buy much more legislation in Nevada.

Nevada is much less hostile to corporate interests than California so the amount of legislation they can buy with via lobbying is only part of the issue. I only mention the budget and employment issues because I believe they are relevant to the use of autonomous vehicles in Las Vegas which I believe the parent poster was implying. That would be seen as killing jobs (even if not true) by the local population and could end up being very bad for the legislators that approved the legislation allowing autonomous vehicles. As far as Nevada's budget deficit goes, the issue could be resolved relatively simply however no one in the legislature is even willing to talk about the steps necessary to do so.

I expect the initial trials will be on desert roads.

“What if I could take out my phone and say, ‘Zipcar, come here,’ ” he asked an industry conference last year, “and a moment later the Zipcar came around the corner?”

So nice to see an organization with so much wealth and power spending it on important problems.

Driverless cars could:

1. Save lives by eliminating drunk, distracted, and sleepy drivers. 2. Greatly increase fuel efficiency. 3. Permit commutes where people can work, sleep, or watch movies, enabling greater freedom to live where they wish. This is more economically efficient by allowing more people to take jobs they otherwise would not have.

Yes, I'm very glad to see Google working on problems that save lives, save the environment, and make everyone happier. I'm glad you're glad too!

Did my terseness come across as sarcasm? I was actually quite sincere, and agree with everything you've said.

Yeah, it did, plus you were downvoted, so it seemed like other people agreed. Sorry.

This technology has the potential to save so many lives. Does anyone know how much these cars cost? I would expect the radar system and servers to be pretty expensive. Presumably the higher cost could be offset by lower insurance rates.

We already have cars that parallel park themselves. I just saw an SUV do it in a commercial last night. I think it was a Ford Exploder. We also have adaptive cruise control on a lot of cars. You combine those technologies, plus with something that keeps you in a lane, you got a car that can drive itself. Autopilot.

That "keeps you in your lane" technology is the last piece of the puzzle, and shouldn't be a tremendous legal hurdle if the other two pieces are already legal.

BMW is not going to like this, but it sounds good for Audi. At least based on their company slogans.

Audi/VW are actively investing in driverless car technology including the lab at Stanford.

Does Nevada have stop lights?

Is that supposed to be a serious question? Of course Nevada has stoplights. There are nearly two million people in the Las Vegas area and around two hundred fifty thousand people in the Reno/Sparks area. Then there are LOTS of smaller towns with populations as low as twenty five up to fifty thousand. Sure there are huge amounts of unpopulated or sparsely populated areas but that's not much different than the central valley in California as an example.

Boom - it wasn't serious! It was meant to spark discussion on how on earth do these cars handle trickier aspects of driving on unmodified grids, such as traffic lights.

I would assume since they've already logged lots of hours driving around CA that the various sensors and cameras can already detect traffic lights. On that same line of thought I'd love to know how they adjust for speed in construction zones with non standard speed limit signs, the randomness of other drivers etc.

The last bit at the end makes me think that Google must have laughed at the Kinect.

Who would be responsible if an accident happens?

Whoever is liable now when an accident happens because of a technical failure. My guess is that insurances will love driverless cars, so it's really not a big deal.

I'm awaiting the first autonomous car to get its driver license. I imagine the day is not far off.

I bet the first time robots are legally recognised as sentient (not saying they are now!) will be in court for a liability claim :oP

who is liable now, though? I recall learning about technical errors leading to some devastating, or potentially devastating problems, but the builders weren't necessarily held liable to the full degree?

I imagine there would have to be some sort of blackbox device like airplanes in for safety, learning and obviously insurance purposes...

I have no clue about the legal situation in the US. I do know that in my native Germany, the driver is always liable (even if not guilty).

Since anyone who owns a car (and wants to drive with the same on a public road) has to have insurance this really is no big deal. Currently accidents can result in you having to pay more for your insurance and different people have to pay different amounts of money (an 18-year-old with a Golf GTI has to pay more) but there really is no rational reason for insurances to do that when someone is driven by a robot car. Insurances are interested in statistics, not fear mongering.

I really don't see the big problem. Compulsory insurances solved the problem of liability without guilt a long time ago (1940 in Germany – hey, it's a Nazi law).

You might argue that the manufacturer of the robot car should pay for the insurance but I don't think such a change would even be necessary. Even if, it wouldn't be a big deal.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact