Google: Our Robot Cars Are Better Drivers Than You

An anonymous reader writes "At a robotics conference in Santa Clara, California, the head of Google's autonomous car project presented results of a study showing that the company's autonomous cars are already safer than human drivers — including trained professionals. 'We're spending less time in near-collision states,' he said. 'In addition to painting a rosy picture of his vehicles' autonomous capabilities, Urmson showed a new dashboard display that his group has developed to help people understand what an autonomous car is doing and when they might want to take over.' This follows another (non-Google) study earlier this week that found the adoption of autonomous cars could save thousands of lives and billions of dollars each year. Urmson also pointed out that determining liability for an accident is much easier using the data collected by the autonomous cars. At one point, a test car was read-ended, and the data showed it smoothly braking to a stop before being struck. 'We don't have to rely on eyewitnesses that can't be trusted as to what happened — we actually have the data. The guy around us wasn't paying enough attention. The data will set you free.'"

Comments

Show time

[mjwalshe]

Have the Google robot take on the Stig round the top gear test track.

Re:Show time

[HxBro]

Some say, "he is actually the robot driving the autonomous cars... All at the same time"

Re:Show time

[ravenlord_hun]

Where I come from, that's called "gross negligence" and "endangering lives". If you need to be taken to a hospital ASAP, call the police or an ambulance. They are trained to drive ignoring the traffic rules, and are equipped to warn the other drivers that someone ignoring the rules is coming. If you had failed to yield (because you're going with 80mph all the time) and killed someone, that'd have been certifiably less cool, huh?

Higher-than-thou attitudes like this are one of the reasons why I would absolutely welcome self-driving cars. And a lot more stricter police enforcement of existing traffic rules.

Re:Show time

[hawguy]

call me when your injured and they want one of these to drive you to the hospital. then tell me how you think of these "autonomous" cars. i'm alive because someone put me in their car as kid and drove me to the hospital as a kid doing 80 the entire drive. believe me these things are going to kill people and the makers are going to be all "it's a flawless system"

Of course, a parent rushing a child to the hospital is not in the best frame of mind for driving, and is more likely to get into an accident, like this:

http://millburn.patch.com/groups/police-and-fire/p/local-girl-hit-by-a-car

So maybe the 2 minutes saved by driving 15 miles to the hospital at 80mph instead of 65 mph isn't worth the risk to others on the road. And definitely the 6 minutes saved by driving 80mph instead of 45mph isn't worth the risk to other drivers from driving nearly twice the speed limit on that 45 mph road. There's a reason why emergency vehicles have those bright flashing lights and sirens - and even emergency vehicles get into accidents while rushing to and emergency.

In most cases, you're going to be better off calling an Advanced Life Support ambulance so the paramedics can evaluate and stabilize you on the way to the hospital, but if you choose to drive there yourself, you're likely going to be safer in a self-driving car that's not going to take unwise risks.

Re:Show time

[wonkey_monkey]

The plural of "anecdote" is not "data".

If you're going to use such specious reasoning, your driver could have driven you into the path of a falling meteorite that an autonomous car would have missed by being slower.

Re:Show time

[rjh]

Where I come from, that's called "gross negligence" and "endangering lives".

I don't know where you are so I can't comment on your local laws. In the United States, it would likely be considered neither. Acts necessary to save human lives are neither criminally prosecutable nor subject to civil litigation.

The important word in that phrase, of course, being "necessary." Here's how a judge would evaluate your affirmative defense of, "Your honor, I had to drive like a madman: I had a man in obvious cardiac distress in the back of my car."

• First, did you have a reasonable belief the person was in extreme cardiac distress? "He was clutching his chest, short of breath, complaining of chest pains and having trouble remaining conscious. Okay, yes: the driver had a reasonable belief this individual was experiencing a life-threatening medical event and timely treatment was necessary."
• Second, was your action reasonable in light of the other options which were immediately available to you? "The defendant didn't bother to call an ambulance... then again, we *are* living in Detroit, where the response time to an emergency call hovers around one hour. His options were to either bring the guy to the hospital in his own vehicle, or attempt to provide cardiac care right there in the apartment. Transportation seems like a reasonable choice."
• Third, were unnecessary risks taken? "Sure, the guy was barrelling down Jefferson Avenue at 80 miles an hour. That was necessary. If he'd taken a detour and gone 80mph down a side street to hit a 7-11 along the way to buy a Slurpee, that would've been unnecessary... but he didn't do that, or anything like that."
• Fourth: if there was a reasonable belief someone's life is in jeopardy, if your action is reasonable in light of the options available to you, and if you avoided unnecessary risks, then brother, you are protected.

I am generally not a fan of urban driving. I own a Mustang GT and I go to the speedway whenever I can to race at high speeds in a controlled environment, but once I'm on public roads I obey the speed limit and I live in mortal fear of Suzy Homemaker in an SUV who's jawing on her cell phone instead of paying attention to her lane merge. I welcome the development of automated driving: for 99% of people it will be a massive step up in safety.

But let's not pretend that driving at 80mph in response to an immediate threat to a human life is something that we need to condemn. Those drivers amount to such a vanishingly small fraction of all accidents that I'm happy to give them a free pass. Go with God, may your tires have good tread, and I hope your passenger makes it.

Re:Show time

[Anonymous Coward]

What a terrible argument against autonomous cars. Are you claiming that if they had only driven 65 mph then you wouldn't be alive? How about 70 mph? Unless you were way way out in the middle of nowhere I find that hard to believe. Protip- instead of the hospital stop at a fire station (if you pass one) instead, as the ambulance can do most things the ER can.

Also, why do you think these cars won't have a manual mode?

Re:Show time

[ravenlord_hun]

ONE HOUR? Holy moly. Here you are supposed to have an ambulance there in 20 minutes tops if you call from a residental area. One hour waiting line... that comes real close to not having any ambulance whatsoever. Seriously, if I can wait one hour, I don't need immediate medical attention!

Interesting points, btw! If the police caught you doing that here, you would have a hard time talking yourself out. You are simply supposed to leave it to the proper authorities.

Re:Show time

[rjh]

Detroit is infamously bad, yeah. 58 minutes is the *official* Detroit response time. A few years ago I had to call the ambulance in Detroit for a neighbor who was having a stroke. We never found out what the response time was. We called the ER, who told us to bring her down ourselves. By the time we took her to the ER, sat with her through her diagnosis and admission and returned home, the ambulance *still* hadn't arrived. So I called 911 and canceled the ambulance call.

Re:Show time

[rjh]

Mostly Detroit having been in a state of slow-motion collapse for 30+ years. Even the bankruptcy is caused by that -- it's not as if it suddenly came out of the blue.

30 years ago Detroit had 1.8 million people. Today it has about 700,000. A lot of businesses have also left, too. The city has spent 30 years acting as if nothing has really changed while the entire tax base has fled. Now the city is in a financial emergency of unthinkable proportions. Something like two-thirds of the ambulances have over 200,000 miles (320,000km) on them; there are 40% fewer police patrolling the streets than there were a decade ago; to save money, the city has shut off streetlights in something like half the city.

To make matters worse, half the city is functionally illiterate and thus can't find work in a modern economy. Unemployment in Detroit hovers around 50%.

Detroit's problems are the result of the city itself collapsing. The bankruptcy is just a symptom of the much bigger problems. Even if the federal government were to cut a $20 billion check to bail Detroit out of bankruptcy, these deeper problems would still exist.

Re:Show time

[canadiannomad]

911: 911 what is your emergency?
10 year old: I think my dad is having a heart attack.
911: How old are you?
10 year old: 10
911: Ok, thanks, we have your address at 123 Flower St. Is that correct?
10 year old: Yea
911: Is your dad's car at home?
10 year old: No it is out with my mom
911: Ok we can route a taxi to your home, it will get there faster then an ambulance, do you think you can help him to the taxi?
10 year old: I think so..
911: It will be there in 40...30 seconds. Don't hang up your cell, in case you need assistance.

Taxi arrives exactly as predicted by the operator, child is able to help their father get in, the taxi detects that both are safely in the car and speeds off to the hospital automatically, not asking for fare. The 911 operator walks child through basic emergency procedures though the taxi communication system. All other cars yield properly as the taxi is sending out an authorized emergency signal and has lights flashing and horn blaring at a recognized interval. They arrive at the hospital in a minimum amount of time with less risk then waiting for the ambulance to arrive.

I'm sure that isn't exactly how it would go down, but I can certainly see the advantages of autonomous cars, taxis, even ambulances at every step of this.

Re:Show time

[amorsen]

It would be completely trivial to let a human override the limits. The car could easily be made to automatically alert authorities and even the other autonomous cars that an emergency override was in effect. That should make the journey considerably safer than when a distressed human driver is trying to deal with an emergency amongst a lot of other traffic who just think he is being an idiot.

Abuse is equally trivial to deal with. Every time you activate the override, you get a nice little chat with either the police or a judge.

In Denmark the law is quite easy: If you are doing emergency driving, you must attach a white piece of cloth to the car if possible (sticking white linen in the bonnet is the typical solution). Other traffic must obey the exact same rules when dealing with a marked car as they would if it was a regular emergency vehicle with the lights flashing. The driver of the marked car is not subject to regular traffic laws (but must of course still try to avoid accidents, it is not a license to kill). Once the emergency is over, the driver must report the journey to the police as soon as reasonably possible.

Re:Show time

[Plunky]

Which us why self driving cars are better for all road users.... They won't drive at triple the speed limit.

I don't get why people think they wouldn't do that

In fact, in a world of self driving cars, I don't see why we would need a speed limit. The car can be trusted to drive within its own limits..

And if your dad had a self driving car, your brother could have helped your dad to the car and told the car to drive to the hospital.

and his brother could have told the car that it was an emergency (destination: emergency unit of hospital) and the car could have decided to drive at its maximum speed, broadcasting to other vehicles that it was doing so.

A journey such as this might attract some official attention, but assuming that the hospital corroborates the story of an emergency, then I don't see why there would be any charges..

Re:Show time

[fisted]

That's a nice way to deal with things.
The only problems is, it makes too much sense. The US are never going to adopt it.

At what speed?

[Big Smirk]

Autonomous cars will more than likely drive at exactly the speed limit. So on that stretch of highway you were used to doing 65mph in a 55 zone... well that slow car (hopefully in the right lane) will be the Google one.

I guess that's when the human takes over?

Re:At what speed?

[beelsebob]

Alternatively... you could just drive within the speed limit?

Re:At what speed?

[mbone]

I can see this as having a raft of unanticipated consequences.

Suppose that driverless cars never broke the speed limit & other traffic laws, except in an emergency. Then, revenue from traffic tickets would disappear. Now, many police departments rely on those revenues. So, will they shrink, or find some other source of revenue? (I suspect the latter, and worry what that might be.) And, both the safety and the revenue desire to keep speed limits low will largely disappear, so many speed limits are likely to rise. Likewise, low speed limits are also used to keep people out of residential areas, and that could be accomplished by setting navigation preferences in the autodriver's GPS system, so those could rise too.

And, of course, if you want to have a mistress, she had better be within walking distance, or accessible by public transportation, lest Google start sending your wife ads for Private Investigators and Divorce Attorneys ...

Re:At what speed?

[JaredOfEuropa]

I think we'll see some high-speed lanes (like diamond lanes) for robo-drivers only. They can do 160km/h safely, bumper to bumper. When there are enough robo-cars, the main reason to impose a speed limit at all is noise or environmental concerns

Re:At what speed?

[ShooterNeo]

The first car cannot avoid the deer, true, but this condition will not cause a pileup.

Think about the physics. The inelastic collision between the deer and the car will marginally slow the front car down, true, but only slightly. (since a car weighs 2000 kgs and a deer weighs less than 100, for an estimate). So the combined car-deer vehicle will be going only slightly slower.

Ok, so now the car that is about to hit the deer applies maximum braking force. It begins to decelerate at a rate limited by friction between ground and car. This friction is independent of the mass of the car, for reasons I can't fit into here.

The moment it hits the brakes, the car behind it will see the distance between the two begin to decrease. They are "bumper to bumper", or within 1 meter of each other. The car behind will apply maximum braking force the very moment a single cycle of it's control loop happens (probably 1/1000 or a second or so).

The car behind that will do the same, and so on.

As long as no car in the pack has significantly better brakes than the other cars, no one will hit anyone. Even if a particular car does have better brakes, the collision will only do slight damage, as the relative velocities will be low.

Contrast this to what can happen in a real highway, where a car in front can have time to decelerate to a stop in some cases, and the cars behind may be driven by a distracted driver who does not see the stopped vehicle in time. The collision happens at highway speeds between the trailing car and the stopped car. This, in some cases, will be fatal.

Re:At what speed?

[Guppy06]

Try it sometime and you'll be passed left and right.

So?

It sounds like autonomous cars are safer because they're not so concerned with "winning."

Re:At what speed?

[beelsebob]

The biggest incorrect assumption here is that going 10% faster will get you there 10% sooner. Not only is the maths wrong, but it ignores that the actual result is that it gets you to the back of the queue at the traffic lights 10% sooner, and through the lights at the exact same time. Even on a freeway generally all it does is gets you to the back of the queue of slow moving cars slightly sooner, whereupon you get out of the queue at basically the exact same time.

Re:At what speed?

[TheLink]

If you lane change rapidly so that you can go fast it may cause other drivers to brake suddenly. That can create a "traffic wave jam" that persists till the rush hour is over or till the "traffic wave" moves to a light/empty traffic are before then.

https://en.wikipedia.org/wiki/Traffic_wave

That said you don't have to speed to cause other drivers to brake suddenly.

Re:At what speed?

[drkim]

revenue from traffic tickets would disappear. Now, many police departments rely on those revenues. So, will they shrink, or find some other source of revenue?

Conversely, city costs would shrink. There is a good deal of tax and ticket revenue money that goes toward special police traffic units, driver instruction, court costs, emergency services for accidents, highway signage, etc. that would decrease dramatically.

It's possible that auto-drive cars could actually save the city costs.

Yup, and it doesn't matter.

[jeffb (2.718)]

We'll soon reach a point where autonomous vehicles are orders of magnitude less likely than human-driven vehicles to have an accident. It won't matter, though; people would rather face a daily one-in-a-million chance of dying due to their own mistake than a daily one-in-a-billion chance of dying due to a machine failure.

Autonomous vehicles will still take over in the end. It's just that this particular rational motive to make it happen won't be contributing very much. So, it'll take longer than it should, and more people will die.

Re:Yup, and it doesn't matter.

[rogueippacket]

Your assertion that autonomous vehicles will take over fails to take into account one of the major reasons we have such a large automotive industry - people like to drive. They like to buy new cars, repair old cars, and do stupid things in fast cars. At most, a car with auto-pilot would be a convenience feature for the daily commute, but so long as people get an adrenaline rush when they put the pedal to the floor, this will not change.

Re:Yup, and it doesn't matter.

[demonlapin]

Very few people, even those who enjoy driving, enjoy more than a tiny fraction of the driving they do. In these situations, I find that the best touchstone is asking what very, very wealthy people do. They have essentially unlimited options, and what they do is reflective of human desire not limited by constraints.

Overwhelmingly, they choose to be driven. They choose to fly private jets. If you could afford it, you would do the same thing most of the time, because most of the time getting there is just a task, not a joy.

It will be the same with regular people. Imagine what society looks like when there are zero deaths due to drunk driving, distracted driving, and falling asleep at the wheel. Imagine how much lower car insurance premiums are when the risk of an at fault accident is nearly zero. People will still buy cars, because they will want one customized to them, but imagine all the things that can change when a human pilot no longer has to be accommodated: cars set up so that parents and children can face each other and play games together while traveling, lay-flat seats for overnight driving. You can leave Washington after work on Friday and eat lunch in New Orleans.

Re:Yup, and it doesn't matter.

[hawguy]

Except, if you've read any of the news for the last few years, it IS changing. Young people aren't entering into the automobile culture the way their parents did. They are favoring bicycles, walking, public transit, and other non-car ways to get around. It's something the big car mfgs are worried about a lot because their customer base is rapidly aging.

http://www.digitaltrends.com/cars/5-reasons-young-people-are-not-buying-cars-or-getting-their-drivers-license/

And as a cyclist, I trust the self-driving cars *much* more than human piloted cars. I see drivers texting while driving every day, and I'm confident that a self-driving car will never be reading a break-up text from its boyfriend and plow into me while txting a reply.

Re:Yup, and it doesn't matter.

[Libertarian001]

Oh, we're going to have this discussion again. Okay, I'll play. You see drivers texting and putting on make-up. I see cyclists riding in the pedestrian crosswalk, riding against the flow of traffic, maneuvering around slowing/stopped vehicles, and they never EVER use hand signals to indicate a turn or stop. And most of them don't use required safety equipment, let alone the recommended ones.

In aviation, more maneuverable aircraft yield right-of-way to less maneuverable aircraft. Though not actually codified, this is generally true in automotive. No one argues with the 18-wheeler. Then it breaks down when cyclists expect everyone to move for them (and this is the exact argument another gentlemen here was making the other day when claiming it's the driver's responsibility to adjust *their* behavior to accommodate cyclists).

Maybe motorcycles and bicycles should also be automated. I mean, fair is fair.

Re:Autopilots

[Sique]

If you had read TFA, you would have noticed that the robot car operates more safely than humans in the highway infrastructure that is in place today. We don't need to redesign today's infrastructure, if we switch over to autonomous cars.

Re:What is the use of being better Driver?

[mbone]

Get it into production, allow for Moore's law, and these could be competitively priced in a very few years.

Perfect Synergy

[some old guy]

Thanks to our dear friends at the NSA, law enforcement will soon have the ability to override the destination selection of autonomous cars and have any driver/passenger they wish promptly delivered to a convenient jail or donut shop.

I love technology!

Re:On the desert roads of Nevada, maybe

[jovius]

Solution: robot children and pedestrians. Anyway, wearable computing will make the garments aware of the surroundings. Trying to cross the road? Your pants know better!

Re:Sounds Good

[OzPeter]

if {collision}
then {arbitrary braking profile}
else {real data}

Burma-shave

Good, good ...

[LordKaT]

Now go and take this out into New York City on 5th avenue at 5pm ET rush hour during the work week.

No, seriously, I want to see how well this car performs in a city where the posted 40mph speed limit oin the Staten Island Expressway is ignored by the vast majority of cops and motorists, the normal speed is about 70mph or so, and people will rear end you out of spite if you go too slow for them.

Then get me the data on how much less it costs to run this car.

Data will get you jailed

[almostadnsguy]

1. Will you still be drving drunk if you have your autonomous car drive you home after a night of drinking? 2. What if you are driving link and ass and rear-end someone, will they be able to use that data against you? What if both people are at fault? 3. Who's going to absorb the liability for these cars when something unexpected breaks? The large automotives are going to drag their feet for years on self-driving cars. Their will need to be a lot of testing in real life before they mass produce any cars.

Taking over during emergency...

[HockeyPuck]

So when you're driving today you're in a state of being aware of the situation and are engaged with the surroundings.

If you're letting the car drive, I highly doubt you're paying that much attention. Why wouldn't I let the car drive and I read, do email, surf the web or turn around and talk to the passengers in the rear seats.

In the event where you need to take an emergency action, it's much easy in the first case to go to heightened state than in the second one. Atleast in the first one you aren't completely surprised by the events you're facing before you.

Think of the case of a gravel truck that has a loose load. If I know there's a truck in front of me, I'm not 100% surprised if some gravel comes out, whereby if i'm reading/emailing and I'm forced to take over to avoid gravel, it's more of a surprise and I'm forced to figure quite a bit more out about the situation before I can act. One could also panic because of the amount of elevated emotion or adrenaline dump that would be taking place since you'd go from "reading iPad" to "dodging gravel".

Re:Taking over during emergency...

[Guppy06]

Think of the case of a gravel truck that has a loose load.

The good driver would apply the brakes, gas and turn the wheel to make sure the gravel passes harmlessly over or under the car.

The better driver would remember that there was still traffic to the sides and behind him and, rather than hoping they all have better reflexes than he does in dealing with his own sudden braking/accelerating/steering, lets the rock chip the windshield, which is later replaced on-site within 30 minutes, with costs covered entirely by his insurer.

The best driver notes the standardized "STAY BACK" sign on the back of the dump truck and actually stays back.

Guess which one the autonomous system does!

Re:On the desert roads of Nevada, maybe

[alexgieg]

There is no tech yet that can anticipate a child about to kick a ball out onto a road, or to see that a pedestrian is about to walk out in front of you without looking first.

Do you really think you 300ms senses are better at detecting 'random_object_in_car_path()' and doing a 'controlled_break( distance( random_object_in_car_path() ) / car_speed() );' than a laser detection system operating at sub-millisecond speeds?

Try it in Britain

[GrahamCox]

Last week I had an American friend over and we were talking about driverless cars, and she said she thought they might work in the USA, but having seen what UK roads are like, she was very skeptical they'd work there, so maybe Google should try it!

For example, many roads in tows date back to roman times, and are too narrow for two-lane traffic. You need to look far ahead and work out when exactly you need to duck into a gap behind a parked car to let oncoming traffic through, and when to go for it when you have right of way so as not to block traffic in either direction. And if a block does occur, will it mount the pavement (sidewalk) to free things up, or know when it's time to back up and give in?

The UK has very few towns laid out in a grid, and most roads are twisty, and narrow, other than motorways. Can a driverless car cope with such terrain? If Google really want to prove their technology is better than a human, let them bring their cars over to the UK. If they work here, I'll be impressed.

A few points on self-driving cars

[Animats]

Having worked on self-driving cars (2005 Grand Challenge), a few points:

The comment about minimizing "near-collision states" is significant. A near-collision state is one where a reasonable variance of the behavior of another vehicle could cause a collision. It's about predicting other-vehicle behavior. That's an important area to study. Aviation people put a lot of effort into minimizing near-misses, and it pays off.

Incidentally, Tesla's announcement that they're starting work on an "autopilot" is them playing catch-up. Audi, BMW, Cadillac, and Ford are already demoing automatic driving systems. It looks like Cadillac will be the first to ship hands-off highway driving, in 2015. All these early systems are highway driving only, although Cadillac includes stop-and-go driving in traffic jams. That's likely to be a very popular feature.

On the sensor side, more progress is needed, and it's coming. That rotating LIDAR contraption on top of Google's self-driving cars is from Velodyne. It's 64 LIDAR units on a spinning turntable. That's a research device, not a production one. There are better ways to do LIDAR, but the cost needs to come down. The approaches used in the Kinect and the XBox One will not work outdoors in bright sunlight. Outdoor LIDAR systems work fine, but they're pulsed, not continuous. For a nanosecond, at one frequency (color) they far outshine the sun. But the total energy per pulse is low, so they're eye-safe. Currently, such devices are very expensive, but that's not for any good reason. It's because some exotic ICs have to be made in tiny quantities.

Radars are getting better, too. A decade ago, in the Grand Challenge, we had to use Eaton VORAD radars, which operate at 24GHz. These could reliably range cars, trucks, and larger bicycles, but not people at long range, or signposts. (Such radars return range, azimuth, and range rate; this isn't a speed gun. I used to have one of these looking out my window at at an intersection, with a display plotting the traffic.) Today's automotive radars are running at 77GHz, with plans to move to 79GHz. There's an effort to standardize on 79GHz internationally. Tripling the frequency, plus applying more compute power to the processing, means that most objects a car might hit are detectable. These radars are getting cheap and small, so a car will have enough of them to provide full-circle data. Long range is needed mostly in front; on the side and in back, much lower power can be used.

A key issue is a high viewpoint. This isn't just about obstacle detection. You also need to profile the road. This was a big deal for the off-road DARPA Grand Challenge, but even on paved roads you need to be able to detect junk on the pavement and potholes. Google has their sensor on top of the roof. This will probably be unacceptable in a production car. I'd go for flash LIDARs at the top corners of the front windows. One possibility is a narrow strip just above the windshield, to contain all the sensors. This is one way to combine vehicle aesthetics and field of view.

Cameras are useful, but computer vision is still kind of dumb. Distance from stereo only works at short ranges, and range rate info from cameras is poor. Digital cameras are so cheap now, so it's tempting to think they can do the whole job. Not yet. Computer vision isn't good enough. Tesla is probably putting too much hope into camera processing. You need cameras to recognize signs, traffic lights, and such. Also, you need multiple sensors because not all objects are visible on all sensors. Radars can't see insulators. Cameras can't see objects with little contrast against the background. LIDARs can't see some materials, such as the charcoal fabric used on many office chairs. Sensor fusion is essential.

Enough for now. This looks quite do-able.