Google's Self-Driving Cars: 300,000 Miles Logged, Not a Single Accident

An anonymous reader writes "The automated cars are slowly building a driving record that's better than that of your average American. From the article: 'Ever since Google began designing its self-driving cars, they've wanted to build cars that go beyond the capabilities of human-piloted vehicles, cars that are much, much safer. When Sebastian Thrun announced the project in 2010, he wrote, "According to the World Health Organization, more than 1.2 million lives are lost every year in road traffic accidents. We believe our technology has the potential to cut that number, perhaps by as much as half." New data indicate that Google's on the right path. Earlier this week the company announced that the self-driving cars have now logged some 300,000 miles and "there hasn't been a single accident under computer control." (The New York Times did note in a 2010 article that a self-driving car was rear-ended while stopped at a traffic light, so Google must not be counting the incidents that were the fault of flawed humans.)'"

Comments

Rear Ended

[sconeu]

The GoogleMobile was behaving properly, and was stopped. It had no possible way to evade the puny human that hit it.

However, after the accident, the GoogleMobile was heard asking another car, "Hey, hot mama, wanna kill all humans?"

Re:Rear Ended

[Anonymous Coward]

We don't know if the Google car could have avoided it. I was in a similar situation one time, and happened to catch a glance of the idiot in my rear view. I cut out ahead of the adjacent lane into the empty crosswalk. The idiot screeched to a stop in what was previously my lane.

There isn't always empty space; but if there is then the Google program should recognize it as available for evasive maneuvers. The Google car will not have a heart to go pitter-patter like mine did; nor a father who turned to me and said, "you're a good driver".

Re:Rear Ended

[norpy]

You are creating a straw man there, 99% of similar situations with human drivers would either have not noticed the exit or not reacted in time.
Additionally you likely broke the law doing what you did and if you caused an accident or ran over a pedestrian because of it you would have been 100% at fault, whereas being shunted by the guy behind you lands 100% of the resposibility on him (unless you stopped too close to a car in front of you).

I would put money on your driving record being way worse than 300k miles accident free. The actual pouplation-wide average is a LOT higher than that, and you are asking for us to give up reducing that number because we can't reduce it to 0.

That's like people saying "Don't build gas power plants to replace coal plants because they still emit CO2", sure it's not perfect but at least it BETTER.

Re:Rear Ended

[Beardydog]

I don't think he's saying we should give up on reducing that number. I think he's saying it may technically have been avoidable in that case. It's not meant to be dismissive. It's meant to be food for thought.

Re:Rear Ended

[Cryacin]

I'm more wondering what it would be like for the driver who actually rear ended a robotic vehicle

Uh, yeah, so I rear ended you. We should exchange insurance details.

I'm sorry, Dave, but I can't do that.

Re:Rear Ended

[Dunbal]

But if everyone went around in automated cars, the point is there would be no "idiot" in your rear view because he also would have been in an automated car; one which would have stopped in time.

Re:Rear Ended

[Chrisq]

I'm more wondering what it would be like for the driver who actually rear ended a robotic vehicle

I imagine that you exchange details with the human in charge, with the full knowledge that there will be a complete 360 degree video of the accident with measurements of speed of both vehicles.

Re:Rear Ended

[Anonymous Coward]

Considering that fact, it would be near impossible to rid the road of all idiots, therefore, it would make sense for the automated car to have some sort of collision avoidance algorithm for rear enders that also doesn't put other drivers or pedestrians at risk.

With 300,000 miles without "a single accident under computer control." it has a collision avoidance algorithm that is far better than yours. Therefore it would make sense that you wouldn't be allowed to manually override a far superior driver.

Re:Rear Ended

[Havenwar]

We had a neighbour when I was little who was obscenely proud of the fact that he had been driving for something like 70-80 years without ever being in a car accident. Even then, old, with deteriorating everything (but a sharp mind) he never had an accident!

Incidentally he had also been living on the family farm all those years, and aside from driving the tractor around the farm (and occasionally breaking shit with it, like one time he misjudged his angle going into the barn and tore half a wall down with his rear tire) he actually only drove into town once a month for supplies. A drive of about 15 minutes on a road where you met another car maybe once every five times you drove it, to get to a town where livestock had the right of way and everybody just kind of crawled around in their vehicles around whatever obstacles might appear, be they sheep, pedestrians, or a ninety year old half blind man driving on the wrong side of the road.

I'm just saying, sometimes good drivers have accidents, and bad drivers avoid them, because of whatever outside reasons govern their reality... rear-ended by an idiot, or spending their entire life driving in a very very safe environment. With that being said of course you are right that there are some good drivers who never cause an accident, and of course these are better drivers than an automated car... And there are drivers so bad that they pull the average right back down again.

In short, your argument is invalid - if we replace all cars with robots that have a better than average record, then the average would rise, even if we never let a good driver touch the controls again. And if ALL cars are automated, they can be patched, they can be linked, and they are over all predictable - which is the main risk in traffic today. Unpredictability. Almost every accident happens because of one of two things - 1) something unpredictable happens. 2) the driver failed to predict something obvious due to ignorance/distraction/narcissism/slashdottism/whatever.

Both of those aspects can be near eliminated by letting machines do the driving.

Re:Rear Ended

[Havenwar]

I used to agree with that, then I realized that since we regularly inspect cars and since it would be easy enough to scan the software of every car that is involved in an accident, this is a crime that wouldn't go unnoticed. I like modding, but I'd probably be cautious if I knew that in an accident situation my crime could be upgraded from "manual operation of vehicle outside of my abilities" or whatever to "reckless endangerment" or in cases where someone was actually killed, it could easily be upgraded from "manslaughter" to "murder", or whatever the terms would be in the appropriate jurisdiction.

Sure, some people would still tamper... and the cops would likely have scanners scanning your software on routine stops, maybe even getting feedback from it remotely.

But you know what'll really kill it? Predictability.

If your car is driven by a program that will under no circumstances go above the speed limit, then if you go above the speed limit it's reasonable suspicion that you've tampered with your program. The car is impounded, and if it's found to be tampered with, lost.

Since the kind of people who really love to mod their cars are also the kind of people who really love their cars, this kind of thing is a fairly easily enforceable and not too harsh punishment that would likely make them think twice. And their hobby isn't outlawed, no, just moved to fenced trackways. If they want to race, feel the speed, mod their cars... then that's not for the streets, that's for the raceways.

Will there still be the one in a million idiot that does something insanely stupid, gets someone killed, and then drools in the courtroom saying he didn't know or that the rules somehow magically don't apply to them? Yes. There will also still be lightning strikes. If we got fenderbenders and carcrashes down to the scarcity of lightning strikes however, or lower... man, imagine the billions of dollars and manhours saved... not even mentioning the deaths and injuries avoided.

Re:Rear Ended

[somersault]

Better doesn't imply that, it just means.. better than whatever state you were in before. When people are sick, we often say "I hope you feel better".

Re:Rear Ended

[Havenwar]

Well, the issue here lies in the unexpected. The unexpected does happen, quite a lot, and unless the cars AI can handle this better than a human driver it is not a better driver. I've driven more than a few laps around the planet so far in my life, and I can tell you that unexpected situations are really common. Proving your other point however, the two near accidents I've been is was once when I was young and dumb and showing off, and once when I was really tired and zoned out for a moment. I recovered in time, but of course an automated car wouldn't have had these problems.

On the other hand an automated car would have likely crashed in a few hundred other situations I've encountered, that after having reacted and averted the danger have left me going "huh, what the FUCK just happened?"

So I believe it's possible to give the fair assessment that a good driver is a better driver than an automated one, simply because that's the nature of driving. Unexpected situations. Once we get to the point where all roadway traffic is automated and controlled in unison, then the unexpected situations will be near null, and thus automation will win... But as they taught us in basic physics... "This is only true in a perfect vacuum, without any outside interference."

Re:Rear Ended

[somersault]

I used to think like that, but then after considering the actual problem, I realised it's perhaps not as complex as we think. Basically to be a "safe" driver all you need to do is avoid bashing into anything, or causing anything to bash into you. Now if the thing went over an oil slick or something then I can see that causing difficulty for the AI, but in other situations what with its cameras and radar and whatnot, I think the AI is better placed to tell when someone or something looks to be getting in its way.

But really, most accidents wouldn't fall into the "unexpected" category for someone who has researched into the most common causes of accidents, and generally doesn't trust other drivers to do the right thing. A good driver is always ready for that idiot to pull out without looking, makes sure that they haven't got their turn signal on by mistake, slows down adequately for blind corners (especially when there are buildings nearby, or a known junction on the bend for example). If Google have done their job right, the car probably expects a lot more accidents than any human, and is ready for them.

Re:Rear Ended

[AmiMoJo]

What about cars with no passengers though? Say there is no parking near your work so you send the car home, then tell it to come get you at 5PM. On the way it has an accident. There will have to be some kind of system in place for notifying the owner and allowing the other person involved in the collision to speak with them. Even if it is just a notice placed somewhere on the car it will have to be standardized.

Re:Rear Ended

[Havenwar]

Okay, unexpected things that doesn't include hitting or being hit by things... Hmm.

At one point I was driving down the road, when the road disappeared.

Sure, it didn't literally disappear, it just so happened that it re-located about 30 feet downwards in an instant due to an unfortunate passionate meeting between a boat and a support pillar holding the bridge I was about to cross up. I guess they really liked each other, or whatever, but all the same I was definitely surprised, there was nothing for me to hit but rock bottom and by the time any computer would have seen that it would have been too late, and nothing was about to hit me.

It's possible their car could handle this situation, but that depends largely on how perfectly they can detect every inch of road. From what I've heard they go mostly by road markings and such, which means that in such a situation they'd have to switch to manual control. Hopefully in time to avoid a nosedive into the amorous steel behemoths below.

Children running into the road is childs play - a clear collision avoidance. How about children jumping over you? There I was driving down a narrow "alleyway" between two walls/brick fences that were just barely high enough that I couldn't see over them, and all of a sudden a bunch of teens on bikes fly over the road over and infront of me. I slowed down, and that was pretty lucky because the last guy didn't make it. Granted that's collision avoiding, but would google's car have been slowing down for things flying above it? Can it tell that this is kids on bikes that might crash, or a ball that might be followed (if it's above you and thus not in your direct path), rather than a dove or flying squirrel with a death wish and a meth addiction?

I could go on, but the point I'm trying to make is simple - being a good driver is not just about not bashing into things and avoiding things bashing into you, it's about avoiding putting yourself in a position where you can't stop. It's about avoiding the scene of the potential accident completely, not about behaving competently once you are already there.

Re:Rear Ended

[neyla]

There probably exist situations where creative problem-solving is beneficial, and you're right that humans are better at that for the immediate future.

But this is largely negated by the short reaction-times needed. These are typically split-second decisions, and human beings aren't universally good at being creative and solving complex problems in a few hundred milliseconds.

And often, a excellent solution, where the execution starts after one second, is inferior to a mediocre solution where execution starts in 50ms.

And what matters, is the average performance. If the automated car has a better reaction than that of a good driver in 95% of the dangerous situations, and a worse reaction in 5% of them, then it still comes out ahead, despite the existence of the 5%.

Also, computers and algorithms improve with time -- often rapidly so, whereas human drivers aren't likely to be all that much better in a decade.

Re:Rear Ended

[Havenwar]

I think you're missing the point, it's not about objective skill, but rather the fact that 80% think they are better than average. In reality 50% is better than average, 50% is worse than average. That means 30% of people at the very least, overestimate their skills. It's called illusory superiority, and you can check it out on wikipedia, it's a basic cognitivie bias. It even quoted the studies I referred to.

"For driving skill, 93% of the US sample and 69% of the Swedish sample put themselves in the top 50% (above the median). For safety, 88% of the US group and 77% of the Swedish sample put themselves in the top 50%"
-- http://en.wikipedia.org/wiki/Illusory_superiority#Driving_ability

You can also read the article on the Dunning-Kruger effect.
"The Dunning–Kruger effect is a cognitive bias in which unskilled individuals suffer from illusory superiority, mistakenly rating their ability much higher than average. This bias is attributed to a metacognitive inability of the unskilled to recognize their mistakes."
-- http://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect

My definition of a good driver is one that doesn't get into accidents, and that can handle an unexpected situation quickly and correctly. Whether they obey traffic laws is more about whether they are a good citizen or not, and that's a different argument, however staying within the traffic laws (i.e. lowering speed around schools) prevent accidents, and is as such a trait of a good driver.

Whether you are indeed a good driver is irrelevant to the fact that a lot of people judge themselves as to be better than they are. It's after all not the individual that matters when we're talking about the group as a whole.

Re:Rear Ended

[wisdom_brewing]

Never say never (or always, every, etc...)

It's comments like that that can drive ignorance and mass retardation.

CO2 emissions are NOT bad. Upsetting the carbon cycle is bad. Almost all living (for a given level of living) things emit CO2.

Re:Rear Ended

[artemis67]

Think of the implications of having an automated driving system... the onboard computer is collecting and analyzing data in real time, and it will likely store that information, at least temporarily. So if a Google car is involved in a crash, a full report will be generated, detailing exactly what happened and liability will be very easy to determine in most cases. "Car A has had a faulty motion sensor on the front bumper that the driver failed to have replaced" or "Car B drove through a red light to hit Car A".

I also think that automated cars will observe all safety rules to the letter... like only driving the speed limit (or slightly below), always maintaining a safe distance behind other vehicles, stopping for yellow lights, and having a generous braking distance. Remember, Google could be held liable if the system is reckless, and they aren't going to want that when human lives are at stake.

I think that making automated, passenger-less cars legal will be a very easy decision for legislatures, and will pass quickly. Like I said, I believe automated cars will err far more on the side of caution, like the most grandmotherly of drivers.

Re:Rear Ended

[Chrisq]

Think of the implications of having an automated driving system... the onboard computer is collecting and analyzing data in real time, and it will likely store that information, at least temporarily. So if a Google car is involved in a crash, a full report will be generated, detailing exactly what happened and liability will be very easy to determine in most cases. "Car A has had a faulty motion sensor on the front bumper that the driver failed to have replaced" or "Car B drove through a red light to hit Car A".

I also think that automated cars will observe all safety rules to the letter... like only driving the speed limit (or slightly below), always maintaining a safe distance behind other vehicles, stopping for yellow lights, and having a generous braking distance. Remember, Google could be held liable if the system is reckless, and they aren't going to want that when human lives are at stake.

I think that making automated, passenger-less cars legal will be a very easy decision for legislatures, and will pass quickly. Like I said, I believe automated cars will err far more on the side of caution, like the most grandmotherly of drivers.

I'm not so sure. For driving on the highway I think it will be fine, but think about some other conditions. How will it deal with passing a horse and rider on a narrow road? How will it deal with coming across another driver in a country lane at a place where there is no room to pass? How will it drive on a road with a shear drop off on one side and a cliff on another? If this is narrow with passing paces will it know to pull in so that it almost touches the cliff to give extra space to the vehicle near the drop-off? In busy commuter traffic will it adjust the "aggressiveness" of pulling out from a side-road to take into account that if you don't pull out quick and accelerate hard you could be waiting until the end of the rush? On a rutted farm track can it work out that you have to drive with one wheel on the centre of the road and the other on the edge to avoid the tractor ruts?

If you have a driver there is always the option to safely pull over or stop and say "manual intervention required", but once you allow completely automatic use with non-drivers or no driver the car has to do something sensible.

Re:Rear Ended

[Havenwar]

Huh. My top two driving rules are

1) Everyone on the road is an idiot.
2) That includes you, sometimes.

I find that mindset has made me slow down and look twice more often, and it's kept me out of a few accidents.

Re:Rear Ended

[Richy_T]

Every CO2 emission, [...] is bad.

I trust you'll be ceasing to make them at the earliest possible opportunity then.

what is the issue???

[FranTaylor]

It's hard to imagine being found at-fault when you are stopped and rear-ended.

There's no shame in being involved in an accident if it's not your fault.

We trust others all around us every day to avoid smashing into us. Even the best drivers get hit.

Re:what is the issue???

[mark-t]

No shame, perhaps... but also no less pain, unfortunately.

Sure, they can offer some monetary compensation, but having been in such an accident and received adequate monetary compensation for all my medical expenses, I can sure as heck tell you that I would have rather not have had the money, and had those two years of my life *without* the back pain.

Re:what is the issue???

[Animats]

It's hard to imagine being found at-fault when you are stopped and rear-ended.

Especially when the self-driving car has full video, lidar, and radar coverage of the entire event. And really good lawyers.

Re:what is the issue???

[__aaltlg1547]

It's hard to imagine being found at-fault when you are stopped and rear-ended.

There's no shame in being involved in an accident if it's not your fault.

We trust others all around us every day to avoid smashing into us. Even the best drivers get hit.

The best drivers DO NOT trust others around them to avoid smashing into us. If we did, we'd be the not-at-fault person in a lot more accidents.

I believe it to be incorrect to compare the GoogeDrive cars to average drivers. They should be compared to professional drivers for two reasons:

• GoogleDrive is an expert system: a computer system designed from the ground up to do only one thing and do it extremely well. One does not expect such expertise of people whose driving is incidental to what they do.
• GoogleDrive will only be affordable (at any time in the next couple decades) to rich people to replace professional, expert drivers. So it has to be better than expert drivers to make the case compelling to people who might actually be able to afford it.

Re:what is the issue???

[geekoid]

" to rich people to replace professional, expert drivers."
Mercedes has a history of being first t utilize new technology. Why you think rich people are professional drivers is beyond me. Hell, most 'professional drivers' are no better then any one else.
The tech will take bout 4-6 years to go from high end, it low mid range. The idea that it will take 'decades' is laughable.

Re:what is the issue???

[knuthin]

The only accident that happened with the self driving car, was when it wasn't being self driven. Just explains your point better.

Re:what is the issue???

[GigaplexNZ]

Why you think rich people are professional drivers is beyond me.

What? No. Rich people hire people to drive them around.

Re:what is the issue???

[TubeSteak]

geekoid has revealed himself as one of the hoi palloi that has no choice but to drive himself around.
Come GigaplexNZ, let us retire to the study for brandy and cigars, so that we may laugh at geekoid behind his back.

Re:what is the issue???

[Dunbal]

GoogleDrive will only be affordable (at any time in the next couple decades) to rich people to replace professional, expert drivers.

Remarkably stupid if Google does this. They'll go the way of the "electric" car and all the other fancy cars targeted at, well, Beverly Hills. A few rich celebs will buy one and pose for the cameras, and then they will be forgotten. No, Google needs to bite the bullet and take an example from Henry Ford. Make pennies on the unit, but make millions and millions of units.

Re:what is the issue???

[dgatwood]

Which suddenly made me realize something. Even if Google only succeeds in getting these things in luxury cars, the accident and near-miss rate is likely to plummet for everyone. :-)

Re:what is the issue???

[Anonymous Coward]

> Draw your own conclusions.

That's easy - you suffer from a very severe case of confirmation bias.

Re:what is the issue???

[timeOday]

The best drivers DO NOT trust others around them to avoid smashing into us. If we did...

Ha ha ha... classic...

Did you know 93% of drivers consider themselves better than average?

Re:what is the issue???

[AliasMarlowe]

It is entirley possible that the 93 % are right, if the worst 7 % are REALLY BAD drivers.

Not even really bad. For instance, if 93% of drivers are 1% above the average in driving skill, then the remaining 7% of drivers only need to be 13.3% below average in driving skill. I've had to avoid drivers who are a lot worse than that. My memory (or confirmation bias) suggests that the worst drivers are found in Audis and invariably have a cellphone stuck to their ear, whether in North America or in Europe.

Re:what is the issue???

[TClevenger]

And I have never been incompetently cut off by someone driving any model of VW, Toyota (unless you count the Scion or Lexus), Nissan, Honda, or just about any other non-luxury car brand. Never.

I was nearly killed on my scooter today by a woman in a Honda Accord who ran a stop sign at 25 mph and turned directly in my path. Anecdote is anecdoty.

Re:what is the issue???

[Firethorn]

I've calced out an autodrive system to be 'worth' around $20k to the 'average' person, with the following assumptions:
1. The system is better than 90% of drivers. It may get into accidents that a human would have avoided, but it avoids accidents that a human would have caused. IE it might get into an accident where it didn't recognize the hazard that would have been very obvious to a human, but avoided an accident where the human wouldn't have been able to react fast enough. Whatever, it's somewhere around that 95th percentile. Easily 'professional driver' range for preventing accidents. Result: 90% savings on insurance costs. ($1k-5k/year, depending on record)
2. The vehicle is driven, on average, 1 hour a day, the system lasts 5 years(pessimistic, most cars last 10+), and the ex-driver values his or her time at $10/hour(works, reads a book, surfs the internet while transiting). ($18k)
3. The system results in mild mpg savings(round up)

How valuable would it be to others? Do you drink outside your house? Average cost of a DUI conviction is like $10k. Have a DUI? $1.3k a year for one of those breathalyzer systems, which is a big pain to start your vehicle with AND during driving. Which you might not need if you get an autodrive vehicle instead. Heck, get them widespread enough and it might be mandated for you.

Right now google's system costs $150k per car, with nearly half($70k) being for the laser lidar system. So it's not worth it to the 'average' individual. But what about for cargo moving? Semi-Tractor Trailers? Raise the cost of a driver to $20/hour(more realistic) and 8 hours a day, and it'd be worth it at $320k. Pay some contractor to fill the truck up when it comes in for fuel.

Besides that, there's lidar systems coming out as cheap as $250 per vehicle, which leaves ~$60k cheaper on other parts to make it worth it - and how much of that equipment cost is because it all has to be had made and assembled to modify a car? Wouldn't it be a lot cheaper if it's a factory installed option?

Re:what is the issue???

[Quillem]

Hoi polloi*

Perhaps it's you who has revealed yourself :)

Re:what is the issue???

[Dr_Barnowl]

As the percentage of auto-drive vehicles on the road increases, the effect on traffic flow will be good as well.

A lot of jams on the motorway are just because people have crappy feedback loops, and don't leave enough braking space, and this causes waves of braking to propagate back through the traffic, gathering magnitude, until you end up with a stop.

Auto-drive cars will both leave sufficient stopping space and given a means of communicating with each other, can brake in perfect synchrony, anticipate each others lane changing and turning manoeuvres, etc.

This will have more of an effect on fuel efficiency and the general throughput of the road network than anything else. The only downside to this is that it will become less easy to successfully argue that you should be working at home...

Re:what is the issue???

[dontmakemethink]

Wise man say, when crossing one-way street, look both ways. There are very many hazards that automated cars undo as well. I read an amazing essay about life with all fully automated vehicles, where you don't own a car, instead you punch your cargo and travel plans into a website, and the appropriate vehicle shows up and takes care of your travel needs. If it's a mile or so to the grocery store, a wagon shows up. If it's to the remote cottage an SUV shows up. If it's to a wedding a limo shows up. If it's to an airport a shuttle bus shows up with room for you and your baggage along with others etc. Think about how much time your car is parked and think about how many fewer automated vehicles it would take to service a large population. MASSIVE CO2 emission reduction, especially if most of them are fully electric, as they could easily recharge themselves automatically. The ramifications are really stupendous.

I can't find a link to the essay (I'm unwinding after a long day and I get 3 hrs sleep before a 17 hr day tomorrow), but I'm sure other /.ers have heard of it.

Re:what is the issue???

[u38cg]

Claim incidences for an individual driver are Poisson distributed, with the parameter itself being Gamma distributed. Obviously these are approximations, but surprisingly good ones.

Re:what is the issue???

[coofercat]

In London we have "car clubs" that do this without the autonomous vehicles. Near my house there's a small car parked in a normal residential parking space. I can walk up to it, open the door and drive it around for a couple of hours. If I need to put fuel in, I can do that via fuel card (or reclaim money if I can't use the fuel card for some reason). If there's anything wrong with the car, or I have an accident, I push the "call" button and talk to an operator. All this by pre-booking online, and paying on my credit card after I return the car to it's normal resting place.

It's an expensive way to replace a car you already own, but it's a cheap way to borrow a car for an hour or two a week. When we get autonomous vehicles, it'll probably become entirely more popular, and different vehicles will arrive at your house, as you describe. Can't wait ;-)

Re:what is the issue???

[timeOday]

If I'd known it was going to get technical I wouldn't have mis-quoted my wikipedia link. Here is what it actually says: " For driving skill, 93% of the US sample and 69% of the Swedish sample put themselves in the top 50% (above the median). "

Interference?

[Anonymous Coward]

I want to know about interference between cars. I've only see one self-driving car tested at a time. If there's hundreds within visual range of each other are their radar and laser sensors going to have much more noise?

The little experience I have with robots is that laser range finders like to bounce off things and skew readings. How do the cars deal with that?

Re:Interference?

[stephanruby]

I want to know about interference between cars. I've only see one self-driving car tested at a time. If there's hundreds within visual range of each other are their radar and laser sensors going to have much more noise?

With hundreds of self-driving cars everywhere, then they may even be able to save on gas by flocking together and save on wind resistance and save on stopping time by sharing gas and Slurpees at high speeds.

And of course, fewer sensors would have to operate that way, only the car in front would have to scan far off in the front, and only the car in the back would have to scan the rear.

But how smart?

[Anonymous Coward]

My understanding is that it cannot read signs, or deal with many types of unusual conditions like detours, nor can deal with a location without maps. Does anyone know about the limits of the Google car?

Re:But how smart?

[peppepz]

I also read that the automatic drive wasn't able to cope with simple situations such as another car coming from the opposite direction in a narrow street, requiring manual intervention. So alongside the triumphant tones, they should also explain how much these cars are really self driving; most car accidents don't occur in straight motorways.

Re:But how smart?

[vux984]

Pretty much this.

Google deliberately avoids the more challenging situations, and a LOT of those miles are highway.

There's a reason insurance rates for someone living in a small town in the country are lower. Right now, google is pretty much "that guy".

That's not to disparage what google has accomplished, but its premature to compare it to the safety record of a downtown urban commuter; driving through rush hour traffic to and from work in a major city daily.

Re:But how smart?

[Thelasko]

Google deliberately avoids the more challenging situations, and a LOT of those miles are highway

I work in the auto industry and have seen miles used as a metric in a number of tests. The problem with using miles as a measurement in an automotive test is that highway miles are inherently different than city miles. A manager always says, "this vehicle needs to last X miles. Get the verification data as quickly as possible." The product validation team immediately goes out and puts X highway miles on the vehicle because highway miles are the quickest way to rack up miles.

Later on in those programs, we usually find some issues when the product operates in a city or low speed environment. I've been pushing for vehicle hours as a replacement every chance I get, but people don't listen...

Impressive, but

[punit_r]

It is indeed an impressive statistic about the number of accidents by the self-driving car of Google. This does prove that their decision making algorithms are good.

However, comparison to humans is probably not fair. Human mind is more prone to giving in to temptation. Exceeding speed limits, violating lane changing rules once in a while to get ahead, talking while driving, texting while driving, getting distracted by the hot chick/dude in the car in the next lane are all errors that humans would routinely make. Some of them would lead to accidents where the erring driver suffers an accident. Some lead to an innocent driver suffering due to the errors of others. It is the latter condition where the Self-Driving car's algorithms appear good --- handling exceptions generated by human drivers, pedestrians and traffic.

Re:Impressive, but

[mark-t]

It's not supposed to be a fair comparison. That's the point. The idea is that cars that are driven by computer can be vastly safer than those driven by people, which will have the desired upshot of reducing accidents.

Re:Impressive, but

[Riceballsan]

What exactly is your point? The idea of comparing it to humans is for the decision of whether it is going to be a good idea long term to start migrating cars to being more CPU driven instead of human driven. Unless you are implying that to make the cars more realistic, self driving cars need to start doing more stupid actions to fairly compeat with humans. The fact is they have this track record on roads driven by normal humans, meaning the transition period where some cars are self driving and some are human driven, and in those cases they still out perform humans. I also see more potential for them to outperform humans in surprise situations. Namely because humans make terrible decisions under panic. When our cars start hydroplaning, the first instinct is to slam on the breaks, yet an automated car could actually make a decision, and base the driving on the exact amount of traction each of the 4 wheels is getting, and make a very educated decision and calculation that no human could ever pull off. Same goes if a pedestrian or a deer jumps in front of the car. An algorithm could in theory almost certainly make a good estimate in a fraction of a second as to whether swerving or breaking or a combination has the highest chance to avoid impact.

how does it handle atypical situations?

[Anonymous Coward]

So far I've never seen an explanation, but all these situations have occurred to me within the last year:

(1) Construction zone, worker standing with a temporary "slow/stop" sign indicating when cars can proceed on a one-lane section shared between both directions alternately.

(2) Baseball rolls out into street in residential area, followed soon by child who was initially invisible behind a parked minivan. I knew ball might be followed by someone, and slowed way down so this wasn't a problem. At normal speed, it would have been.

(3) Nearly invisible ice around curve, one other car had slid off road. I knew to greatly reduce speed even below normal winter operating conditions.

(4) Two lanes in each direction road. Noticed other car weaving around unpredictably, and later noticed driver occupied with cell phone. I then knew not to drive next to this vehicle even though that would have been fine in other conditions.

How would google's car handle these situations?

Re:how does it handle atypical situations?

[Anonymous Coward]

I have seen a roadworks site where the speed limit sign showed 0 kph. There was no one working there at the time. I did wonder what Google's car would have done.

Re:how does it handle atypical situations?

[medlefsen]

Here is the interesting thing in my eyes. My guess is that while most of those are actually pretty solvable it is almost certain that there will be situations that trips up the computer whereas a human would have no problems. Even still, the comparison isn't between computers and humans in particular situations, but computers and humans overall.

Imagine if we lived in a world with computer driven cars and someone suggested humans start driving themselves. Imagine the itemized lists people would create that described all the ways that humans slowness, distractibility, and limited data collection capabilities would lead to danger. Humans and computers are good at different things. The issue is which is better on average.

The problem remains though that even if it is safer overall to let the computer drive, are people really going to give up control when, inevitably, there are highly publicized cases of humans getting injured or killed in situations that a human driver could have easily avoid?

Re:how does it handle atypical situations?

[Havenwar]

But we will. That's the entire point... Computer driven cars are better than humans, on average, and in an infrastructure re-modelled to suit such cars they are close to infallible compared to human drivers. We are struggling right now to get a computer to navigate the human infrastructure, but once this sort of machine has saturated the market the infrastructure WILL change.

You have to remember you don't live in the world of yesterday, you live in the world of now. The world of now has a very special aspect to it... what we choose to do, changes the world of tomorrow.

Don't arbitrarily limit tomorrow based on the world we had yesterday.

Re:how does it handle atypical situations?

[sFurbo]

I have a much harder one for you, that I've dealt with myself once (and admittedly not with 100% success): a road covered in fresh snow. No lines visible, everything white. In the dark, with the edge of the road only detectable by a small drop in the snow level and the occasional pole sticking out, snowed-over reflector optional.

Have one of the systems see in a wavelength where the refractive index of water ice is close to the refractive index of air. Snow will be transparent.

Re:how does it handle atypical situations?

[Tom]

how does it handle atypical situations?

By driving with a human inside who can be notified and take over when the robot is confused. Autonomous driving does not mean that the car is out on its own.

Next question?

Re:how does it handle atypical situations?

[Sarten-X]

What if you're in the middle of nowhere and it is simply some debris that rolls over in your path is detected as obstacle. Now you're on the highway doing 65 and your car suddenly slows down/stops for no reason and you get rear-ended.

That's the fault of the idiot tailgating you. If you're driving close enough to the car in front of you that you can't come to a full stop if he does, you're too damned close!

Also, where can I find this awesome magical computer vision & depth perception equipment that can constantly scan 300 or so feet (average stopping distance at highway speeds) and accurately (well lets say >90% confidence) identify any random object in the real world.

The hardware is available from most robot suppliers (being just a few LIDAR units and high-res cameras) and the software can be found within Google's well-secured vaults, being most likely an edge-detection algorithm applied to the pictures, then those shapes projected onto the 3D map from the LIDAR to identify objects, then a pattern-recognition engine to identify the object and its risk. If the object can't be identified with high confidence, assume it's dangerous and slow down until a more accurate assessment can be made.

Re:300k miles isn't much.

[Anonymous Coward]

That's a 1 in 6,500 chance of *dying* in a traffic accident.

Re:300k miles isn't much.

[Cimexus]

Holy crap, it's that high?

I'm never driving again!! :o

But god help you...

[dyingtolive]

The moment even one accident does occur, no matter how mild the consequences or much more unlikely the circumstances compared to a human driver, hordes upon hordes of American luddites will man the lines to do their civic duty to shit upon the idea of cars that drive themselves.

Mind you, this is being said by an American who owns a US made car.

Re:It's not stop-and-go traffic, it's AT&T's n

[geekoid]

lessons that will be programmed in, so no learning curve.
Driving on the snow is simple if you follow simple rules.
Mercedes has a car that can follow other cars automatically, and stop when they do and it works in the snow.

Many questions arise

[qbitslayer]

-What will Google's car do if it gets a flat tire on the road?
-What will it do in case of an accident?
-Can it back itself into the garage?
-Can it parallel park?
-Can it park itself at a commercial parking lot or structure?
-Can it go through alleys?
-Can it go where there are no roads?
-Does it have to have a human on board?
-Can I call it on my cell phone and tell it to pick me up at the airport?
-Can vision-impaired grandma take it for a visit the doctor?
-Can the kids use it to go to school?

There are more but you get the picture.

Re:Many questions arise

[cheater512]

Your questions about whether it needs a human and if you can use your phone to request your car comes and picks you up: Soon.

Google has golf carts doing just that to drive people around their HQ.
You book it online or via your phone and it shows up outside your office, where you can either drive it yourself or let it take you somewhere.

Re:300k miles isn't much.

[geekoid]

That's dying in an accident.
I know you live to hate Google, but at least get you fucking facts straight, you Luddite.

there where 5.8 million reports accidents in 2008
30% chance of being in a serious accident
most accident involve at elast on drunk person.

So the automated vehicle is statistically better in every category.
Frankly, even if the accident had been the vehicle running a red light and smacking into someone it would still be statistically better.

IN short, if you drove or 23 years, during that tine you would have been in an accident. Since i'm a better person then you, I wouldn't hold you being rear-ended while stopped at an intersect against you.

I love this idea

[EGSonikku]

I live in Reno, and Google's Self Driving cars are legal on road here (complete with cool plates with infinity logo: http://www.jumpthecurve.net/wp-content/uploads/2012/05/18164996_BG1.jpg)

A few things:

1) Has google partnered at all with any manufacturers to have this ability on a future car I can buy?
2) or as an upgrade to existing cars?

I'm hoping they don't get stuck in red tape legal limbo hell, and that more states other than my own Nevada jump on board. I regularly make 3.5 - 4 hour drive to friends in California. If I could just jump in the car, pop in an address, and take a nap, play on my iPad, or whatever while the car drove that'd be awesome. Or a ride home from a bar if I've been drinking and don't want to taxi and leave the car behind.

Or imagine a friend asks for a ride someplace? No problem, I send the car over on its own, and he can just tell it to come back to my house afterwards.

There are tons of ideas I can think of where this would be very damned useful.

Re:I love this idea

[RivenAleem]

I've been warned against googling "google". I'm told it breaks the internet.

Google Insurance

[Xaer0cool]

One argument against driverless cars I often hear is that it will never happen because the liability is too great - ie. if someone ran over a baby in a Google car, Google would get sued into oblivion. I think the obvious answer to this is that Google would insure all of it's cars. There is no doubt that driverless cars will be safer, so google could require that to use their driverless car you must have insurance through Google, at comparable rates to other insurance companies. Since Googles car's will be involved in far fewer accidents, the consumer will be paying the same, but Google will be paying out less, so for the odd freak accident, the higher payout due to 'oh nohs teh ebil Google killed my babby!' will be covered because of the lower rate of accidents.

Re:It's not stop-and-go traffic, it's AT&T's n

[Bert64]

Possibly not even for navigation, if the car can recognise junctions, knows where it started, knows how far its travelled and in what direction it can easily calculate this data alongside a map to work out where it is, and where it needs to go.

Re:I wonder...

[YttriumOxide]

I hate it when dickheads like you own nice cars...

Reminded of Schlock Mercenary with this...

[Firethorn]

While I agree in the near term, in the long term I'm reminded of This.
  - While it may seem harsh that the 31st-century equivalent of "Driving Under the Influence" carries with it the death penalty, this is due to an inherent inequivalency between MOUI and DUI.

With DUI, you need only climb into your vehicle while under the influence of alchohol or drugs and attempt to drive it home.

With MOUI you must disable a number of safety systems designed to prevent idiots like you from manually operating their vehicles while inebriated, overtired, wasted, decaffeinated, angry, emotionally distraught, or suffering from hormonal disorders like PMS or testosterone poisoning (the latter having been positively identified as a leading cause of stupidity among males between the ages of puberty and death). After disabling the safety systems (which task almost certainly requires ice-cold sobriety), you must decide to switch the vehicle to a manual mode of operation. In some cases, this requires installing a manual mode of operation.

Other examples would include 'johnny taxi' in some movies. You don't NEED to have manual operation modes once you reach a certain sophistication, worst case you have a sort of protected mode 'guided direction' where you provide steering information - but the car still worries about avoiding accidents, and will override you to do so.

Manual driven vehicles would be restriction to 'special hazard' zones and conditions where they just haven't programmed a vehicle to be able to avoid all the hazards yet. Perhaps a dock loading zone where you have to worry about something being dropped on you from overhead.

Data only if you're at fault

[jeko]

with the full knowledge that there will be a complete 360 degree video of the accident with measurements of speed of both vehicles.

Only if YOU caused the accident. It's a pretty safe bet that if a glitch in their programming caused the accident, there'll be a tragic loss of data... :-)

How many parking spaces?

[david.emery]

How many times has the Google-mobile pulled into and out of parking spaces at busy malls? Frankly, that's where I've had my accidents.

Re:Drive in India...

[Zubinix]

When Google are testing their cars in India then we know they are about ready...

LIDAR cost?

[Firethorn]

Your point? The price figures for the LIDAR was right out of the USA Today article I quoted. Google paid $70k a pop for the LIDAR systems it put into it's cars. There's an unnamed company getting ready to produce LIDAR for cars at a 'mere' $250 each. You quote $30 each, but that's for systems mounted to vacuum cleaners - don't need the range or operating environment tolerances of a car. Besides, your Hizook article is NOT for a LIDAR system, it's for a 'laser rangefinder', which is sort of like half of a LIDAR. Actual LIDAR attempts to build an image, a laser rangefinder doesn't.

At $150k overall, reducing a $70k expense to $250 would make me concentrate more on the rest of the components. When the goal is $20k overall cost(or less), you wouldn't get there even if you got the LIDAR for free. I wouldn't refuse a $30 one, of course.

Though yes, going from hand manufacture and assembly to mass production can save oodles of money per unit.

Rush Hour?

[Bensam123]

I would really question how these cars function in rush hour in a big city. Driving there is sketchy at best and in order to merge into another lane you sometimes literally need to start heading into the other lane even with traffic that isn't helping you merge. How would a car like this function bumper to bumper?

In the future a bunch of these could eliminate traffic jams, but that isn't going to be a case for a long time.

I wish

[SpaghettiPattern]

I wish to be driven about in a self-driving car. For hundreds of kms at hundreds of kmhs. In tightly packed convoys to save fuel even at mind boggling speeds. Sleeping comfortably in safety.

Except for the weekends. Then I wish to exhibit my driving prowess on mountain passes.

Life gets sweeter by the day.

Re:More cars on the streets cannot be good

[gnasher719]

The density of humans per car is already too low. I live in Mexico City, a 25 million people city. There has long been a campain to reduce the use of single-driver cars, but the campain is never strong enough.

I want to go to point X. Being too lazy to walk ten metres, I drive around for half an hour until I find a parking spot. With a self driving car, the car can drop me off exactly where I want to go and leave the area going somewhere where parking spaces are easier to find. Self driving cars can also park closely together because they don't need to leave space on the side for the driver to enter, and they can park blocking other self driving cars, because when the other car needs to leave, the blocking car can get out of the way.