"Fully Self-Driving Cars Are Here" - Waymo To Begin Testing Driver-Free Autonomous Taxis In Phoenix

Tyler Durden's picture

From here on out, if you see a car without a driver meandering around suburban Phoenix, don’t be alarmed: It’s just Google's Waymo division testing its new driverless taxis - the first of their kind to be tested on US roads without the supervision of a “safety driver."

Wayno has revealed that - effective immediately - it will begin testing the driverless taxis - referred to in technologist parlance as a “level 5” driverless vehicle - in Chandler, Arizona. Thew news represents an important milestone that establishes Waymo as the leader in automated driving technology. Waymo CEO John Krafcik made the announcement Tuesday during in a speech at a web summit in Lisbon, Portugal.

“We want the experience of traveling with Waymo to be routine, so you want to use our driver for your everyday needs,” John Krafcik, Waymo’s chief executive officer, said at the Web Summit conference in Portugal. “Fully self-driving cars are here."

According to Ars Technica, for the last year, Waymo has offered free taxi rides to ordinary people who live near the Phoenix suburb of Chandler. Until recently, the company's modified Chrysler Pacifica minivans had a Waymo employee in the driver's seat ready to take control if the car malfunctioned.

One reason the company is so confident in its techonology, as Bloomberg points out, is the Alphabet subsidiary has racked up more autonomous test miles on roads than others developing the tech, including Ford Motor Co., General Motors’ Cruise Automation unit and Uber. However, Google’s rivals have certain advantages that may ultimately help them beat Waymo to market. For example, Uber has a massive customer network that depends on its drivers for rides. And Ford and GM have the manufacturing capabilities to crank out new units with very little delay.

And by the looks of it, Waymo is gearing up to challenge Uber by using its service to begin offering rides. Krafcik, a former Ford executive, said that an on-demand service would be the first commercial use case for Waymo. During his appearance at the summit in Lisbon, he also discussed how the vehicles may replace personal car ownership, a nightmare scenario for OEMs like Ford and GM.

“Because you’re accessing vehicles rather than owning, in the future, you could choose from an entire fleet of vehicle options that are tailored to each trip you want to make,” Krafcik said, according to a transcript of his remarks. People could claim the cars for a day, a week “or even longer,” he said. He ticked off the ways driverless cars could be redesigned if the vehicle didn’t need space for a driver: to ferry groceries, as a “personal dining room” or for naps.

Waymo began testing its taxi service in Phoenix back in April. It’s progress shows how Google has played to its strengths by building the best technology available. However, the question of scalability still remains.

As the New York Times points out, driverless cars are regulated by a patchwork of state laws. Arizona, like many states, has no restrictions against operating an autonomous vehicle without a person in the driver’s seat. On the other hand, California, where Waymo is headquartered, requires any self-driving car to have a safety driver sitting in the front.

However, just because Waymo can legally test these cars, doesn’t necessarily mean they’ve been optimized for safety. In December, Waymo published a report for California’s Department of Motor Vehicles about how frequently its driverless cars “disengaged” because of a system failure or safety risk and forcing a human driver to take over. In the report, Waymo said this happened once every 5,000 miles the cars drove in 2016, compared with once every 1,250 miles in 2015. While that’s certainly an improvement, these types of incidents are hardly rare.

So, the question is, how will Waymo handle these situations when they inevitably start cropping up (indeed, if they haven’t already)? We imagine given all the publicity around several high profile cases of deadly car accidents involving Tesla’s autopilot software, that the company has planned for these risks - or at least, we hope they have.

And we’re not the only ones. Consumer Watchdog, a frequent critic of Alphabet, said that data demonstrated that the cars are not ready to drive without any human intervention and that Waymo was following the Silicon Valley model of “beta testing” a new technology on the public - to a potentially dangerous end.

“It’s the wrong approach when you’re dealing with self-driving cars,” said John M. Simpson, a director at Consumer Watchdog. “When things go wrong with a robot car, you kill people."

To be sure, researchers believe self-driving cars can be safer than cars operated by human drivers because they are programmed to adhere strictly to traffic laws, they don’t get distracted, and they don’t take unnecessary risks.

But that reality is a long way off.

Then again, who are we - the public - to stand in the way of progress? The tech gods of Silicon Valley have spoken, and they’ve said we will have autonomous vehicles commercially available by 2025 - which is ludicrously soon, considering where the technology is right now. Because the reality is this technology needs to function flawlessly by the time it’s put in the hands of the consumer.

Of course, regardless of the cost in lives and damage, once the technology is ready, the world will understand that it was all worth it.

In a bit about driverless cars, Stephen Colbert once joked that they’re “a high tech alternative to dropping a brick on the gas peddle and jumping in the back seat.”

Given Waymo’s safety record. That bit might prove eerily prescient.
 

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
hedgeless_horseman's picture

 

You're in a Johnny cab!

hedgeless_horseman's picture

 

Mars?

We first need to land a man on our moon and return him safely to Earth. 

Right, Elon?

http://www.zerohedge.com/news/2017-07-21/did-elon-musk-just-confirm-moon...

Greenspazm's picture

Elon has the tech: battery powered rockets.

skbull44's picture

Driverless cars. Thank you for solving the world’s various dilemmas. I can sleep much easier now!

https://olduvai.ca

toady's picture

Waymo has been running radio spots in Phoenix about how driverless cars "never get tired", "never get lost" and "never scream "Allah Akbar!" while running down pedestrians".

peddling-fiction's picture

How about "AI" "man-slaughter".

Ooops...

Bubba Rum Das's picture

So, since they will be driverless taxis, one can just walk by the pickup line w/ a cane that has a nail in the end of it, puncturing the tires as you go!

What Fun!

Automatic Choke's picture

I wanna see them drive on snow.

I wanna see them drive when rain coming down in sheets.

I wanna see them drive with sun in their "eyes".

I wanna see them drive through a messy construction zone.

I wanna see them obey a traffic cop waving people to stop and go when the lights are out.

I wanna see them merge into 70 mph bumper-to-bumper traffic on LA freeways at rush hour.

 

That video is pure shit.  It is easy to drive in empty residential streets in a dead-quiet residential community at noon on a clear day after the streets have all been swept.  Now what about the rest of the world?  Let's see them drive through Beruit.  Let's see them drive through New York.  Let's see them drive through Detroit.

OverTheHedge's picture

The temptation to break of these is going to be hard to resist for some people. I wonder if people will take potshots at empty, driverless cars pootling about the countryside. I wonder if people will want to get into a car with bullet holes. I wonder if people will want to get into a driverless car that has been tagged with some gratuitous graffiti, suggesting the ladies inside will negotiate for 15 minutes of knee-trembling, or other delights.

I wonder if this will ever take off (as in the Jetsons?), pardon the pun. I wonder about the disconnect between silicon valley, and real people, in all their smelly, warty, uncouth glory.

SofaPapa's picture

"To be sure, researchers believe self-driving cars can be safer than cars operated by human drivers because they are programmed to adhere strictly to traffic laws, they don’t get distracted, and they don’t take unnecessary risks."

This is precisely why driverless cars are less safe than cars with drivers.  They are inflexible.  Every driver of more than a few hours experience has been in situations where you cannot stay within the bounds of "strict" adherence to the norms and avoid an accident.  You avoid the accident by swerving out of the way of whatever unpredictable event occurs - e.g. the classic moron ahead of you cuts you off.

So let's take this to the extreme.  Let's assume that all cars on the road were driverless.  There is the inevitable software glitch in one of them, and it does something unpredictable.  Even with no "human error", all those cars are now going to have to break "strict traffic laws" or hit something.  Which choice do they make?

The Silicon Valley pinheads are proud that their cars don't screw up more than once in 5k miles.  Jesus!  If I had a screwup every 5k miles, my insurance would be prohibitive.  In other words, I have a way better safety record than this technology.  And I'm no perfect driver.  There are plenty better than me.

This technology has been a mistake from the beginning and continues to be an increasingly dangerous mistake because it assumes that "technology" has a lower error rate than human interaction.  I seriously doubt that premise.

RockySpears's picture

"Even with no "human error", all those cars are now going to have to break "strict traffic laws" or hit something."

 

How do you figure that?  If they always leave stopping distance appropriate to their speed, then it matters not what the car in front does.  Does it?

 

Not that I am in favour of them, just do not follow your scenario,

 

RS

SofaPapa's picture

Only twice have I had trouble with the car ahead of me, and when I did, it was because I was an idiot and as you point out, I didn't leave enough stopping distance.  Been there, done that, fortunately, not in a major way, meaning I ended up with a low speed rear-end collision.  I'm quite ready to concede that these would decrease in driverless cars.

It's the side-to-side that's tricky.  Suddenly, even "non-traffic" items get involved.  Deer, for example.  I live in a metropolitan area, and even here we still have a buttload of deer around.  Much more seriously, what about when a little kid gets away from his mom and runs into the street at less than the cars stopping distance?  Try to predict that one.  Hit the kid?  Hit the other cars?  And this is just two examples.  My general point is that driving - for all that it appears so simple and straightforward - is actually a response to an extremely large number of inputs, and a large number among that large number are inherently unpredictable.  I still believe the "human computer" (the brain) is a better tool for handling this kind of randomness than are the codes these tech lovers are writing.

Time will tell.

cheka's picture

and if you dont like it, they'll launch as many terrist attacks as it takes

Ajas's picture

Just remember that car crashes kill 35,000 people each year in the US, with another 2.5 million injured.  Unfortunately, we humans on the aggregate haven't set a very high bar for AI to leap over.

Calm, sober, and alert humans will be better drivers for a long time... but if some ass-hat wants to text or drink or road rage around or drive on the verge of dozing off, we're all safer taking the wheel out of their hands.

SofaPapa's picture

This is an interesting question.  No doubt the impaired driver is less safe on the road than AI, but the more important question to me is which do we want in the cars around the impaired driver.  As one of the other commenters noted, there's a lot of human judgment that is not easily "coded" when interpreting driving behavior around us.  Other drivers see the impaired driver and react accordingly.  What does an AI algorithm "learn" in this situation?  Just one of an almost infinity of precisely the impossible to predict (and therefore code for) variables that go into a realistic road environment.  I'm not convinced "code" will do this better than we do.  In fact, my instinct is that it would do worse.

Two last points:

1) To be fair, I do think it would be possible for a substantially safe (>>99% safe) algorithm to be developed, so the car will not get into any accidents.  In that case, however, I'm not convinced the car would ever get anywhere, because if the machine "learns" to react with pure safety to every circumstance, then the regular road environment will provide so many triggers to it, it will take an hour to make a 10-minute trip in rush-hour traffic.

2) Among the most complex technology we have developed so far is the internet.  Then, in some cases, they have tried AI on the internet.  What is the error rate with these two technologies in each of our personal experience (i.e. pages screw up, networks go down, etc)?  Is this the error rate we want when our lives are on the line in our car, where our personal safety would be the responsibility of the machine?

Driving is complicated, not nearly as mechanical as we assume it is.  Even if 99% of the time, it's routine, that other 1% is why we drive cars.  Cars are not able to drive us.

OverTheHedge's picture

In think it will be fairly easy to assess: the average distance travelled per accident will be easy to calculate. For humans in the US, 12.5 deaths per BILLION miles travelled, allegedly (https://en.m.wikipedia.org/wiki/Transportation_safety_in_the_United_States). As long as SkyNet can do all least as well, then driverless cars are here to stay. Of course the statistics are new, and could be all over the place to begin with.

Urban Roman's picture

Every 5k miles? They need to be class-action sued into oblivion by every human driver. If there are legal barriers to that, tar and feathers.

What makes me nervous is that these drivers are not human. When I am driving around the neighborhood, I at least have some idea what the other drivers are thinking. If they are frantically waving their hands, flipping me off, honking, or perhaps making a polite gesture that I may proceed. I have no way of knowing what a mechanical intelligence is thinking or planning to do.

For one recent example, I was sharing the road at night with a car whose headlights were off. I knew to stay out of its way, because that driver is either crazy, drunk, or has a malfunctioning car. Does the AI car know to do that? What if some of its self-driving sensors are not working, or get splattered with mud? Does it just stop in the middle of the highway? Can it still figure out where to safely pull off (and not crash into the ravine or plow into barricades)?

Yukon Cornholius's picture

Or what if Russian yahoo hackers hack the driving and safety software? Seriously. What if?

Trogdor's picture

Despite the hardware/software failure issues, I'm wondering how they're going to manage to keep them clean. Say the guy or gal before you was sick or really drunk and pukes all over the interior. Or some lady with a baby with explosive diarrhea has an accident, or just someone who spills their 99oz Cherry Slushee all over the seats? Are they planning on having a depot where the cars will go when they detect their uninhabitable? Additionally, the idea of hot-bunking (CEO's comment about naps) was kind of creepy. In the military, hot-bunking is unsettling enough ... and that's under controlled conditions. The idea you could have a car "drive you around while you take a nap" - not knowing what kind of cooties the previous occupant had is NOT appealing at all.

MEFOBILLS's picture

 Jesus!  If I had a screwup every 5k miles, my insurance would be prohibitive.

It is a big problem, but he technology is maturing rapidly.  Waymos Lidar is at 1550nM, so it can see things with pretty high resolution.  Then there is the new Luminar technology that can see even further. 

"Processing vast amounts of sensor data instantaneously and providing enough computing power for autonomous vehicles to determine the correct course of action as better than a human is a substantial challenge." 

Waymo is using Intel inside, and other's are using Nvidia:

https://www.forbes.com/sites/alanohnsman/2017/09/18/when-it-comes-to-way...

As an aside it was ((Levandowski)) who stole secrets from ((Google)) to give to Uber. 

https://www.theguardian.com/technology/2017/may/30/uber-fires-self-drivi...

Uber in turn is part owned by  Travis C. Kalanick born to a Jewish mother (her maiden name is Horwitz) in Los Angeles.

http://jewishbusinessnews.com/2015/01/01/jewish-owned-uber-expelled-from...

It's hard to keep up with maneuvering of tribesters, even when they are busy back-stabbing each other.

Wang Dang SP's picture

They're saving that for the export model.

SafelyGraze's picture

cars, like guns, can be used as deadly weapons

they should be kept away from the control of private citizens, who sometimes abuse their privilege to use them

the main benefit of autonomous cars/weapons is that nobody commits a crime when people are killed

the downside, of course, is that there are fewer opportinuties for prosecution, i.e., plea-bargaining

hugs,
the association of plea-bargaining prosecutors of deadly formerly-criminal acts

Endgame Napoleon's picture

Only problem: There will be more out-of-work, underemployed and, in a few cases, disgruntled males, formerly in truck driving, which pays some males a living wage. These males cannot just have sex, reproduce and get their major household bills paid by government, with a $6,269 child tax credit on top, for adding more humans to the planet to compete with robots for jobs. For all of the humans living on earned-only income that is not augmented by unearned income for womb productivity, driverless cars could help eliminate the multi-layered expense of car ownership, which, like rent for the non-pay-per-birth crowd, is increasingly unaffordable. Those driverless car rides better be cheap, though.

not dead yet's picture

You might be able to get away with driverless interstate for trucks but what about the rest of the country. Do we map every square inch so the trucks can back themselves into the customers dock? Or navigate the customers messy disorganized property. Have multiple cameras inside the trailer so the customer doesn't steal the other customers products necessitating constant monitoring at home base. Or when hijackers block the front and rear and steal the loads does the truck call police who could care less? Or customers could care less employees who just dump their stuff in the trailer where it ends up damaged and wasted trailer space raising shipping costs. A driver does more than just drive the truck. When the driver picks up a load he signs the bill certifying he received the goods in good condition and verified the count, any damage or missing items the trucker is on the hook, or signs for a load sealed by the customer. As long as the seal is intact it's the customer who eats any losses or damage otherwise the trucker does. The driver makes sure multiple shipments on a single trailer are loaded to prevent damage and maximize trailer usage.

Most of your trucks and trailers have swing open doors and stick shifts which would require for driverless all new trucks and trailers to automatics and roll up doors not just retrofits. If an oufit were to go driver less it would require huge outlays in a business where margins are razor thin. Costs would be even higher if they switched to electrics. They would sell off their existing fleets for cheap which would end up in the hands of independents who would undercut and take away business from the adoptors. Eventually there may be driverless fleets of trucks but it won't be in the the near future.

RockySpears's picture

Replace the Driver with a Min. wage employee.  Someone to open doors, watch the stock etc.  

 

It is not always about eliminating jobs, but reducing the cost of a job,

 

RS

tmosley's picture

Love how the peanutz were saying that this would "never happen" as recently as last week.

Juggernaut x2's picture

Autonomous automobiles will eventually take the serfs directly to their jobs and directly home- maybe your overlords will allow you to occasionally stop and spend company scrip- aka Bitcoin- at the company store 

TuPhat's picture

It hasn't happened yet.  It's only a beta test.

UselessEater's picture

UN Agenda 21 baby

http://www.renegadetribune.com/un-agenda-21-preparing-america-poverty/

 

note comment by Anthony Roberts:

Electric light and heat.
Electric cars.
Electronic cell towers.
Electronic cashless society.
Electronic vaccine chips for children.
Electric slaves.
If we don’t fight back now, the jews will soon control all the cattle prods.

Endgame Napoleon's picture

Well, driving is increasingly unpleasant, reducing the feeling of independence associated with car ownership due to endless, clogged traffic lines. As we add legal and illegal immigrants by the millions—paying many of their household expenses and issuing child tax credits in increasing amounts with each child they produce to compete for automated jobs, and doing the same thing for citizen single moms who work low-wage jobs at the welfare-reform minimum of 20 hours per week—the roads will become even more gummed up with slow-moving traffic.

VWAndy's picture

 I called it right. They rolled out the red carpet for this pile o stupid. Red tape? Nope. A free pass on all of it.

Kidbuck's picture

We bought a used Buick in Phoenix a few years ago. Had the wife drive it home through Chandler, AZ. On the freeway at 70 mph a crew of illegal wetbacks kept cutting in front of her trying to cause an accident. They nearly succeeded. Can't wait to see how a driverless car is taken advantage of in these insurance scams. Also, how hard will it be to just get behind the wheel and steal these cars?

tmosley's picture

Cameras will result in 100% of such scammers going to jail.

Theft is pointless, as the thing has GPS. It's like trying to steal a piece of land.

Endgame Napoleon's picture

It’ll be harder for human smugglers in the era of self-driving cars. If they try to do it by air, they’ll have to take off their shoes and go through body scanning, compliments of TSA. We see you have 7 other humans, hiding them under the ornamentation on that Christmas sweater, ma’am. Please step aside.

BandGap's picture

I am sure they are safe and could never be hacked. /s

krispkritter's picture

I should start writing a 'dyslexia' virus for them right now...

mkkby's picture

"In December, Waymo published a report for California’s Department of Motor Vehicles about how frequently its driverless cars “disengaged” because of a system failure or safety risk and forcing a human driver to take over."

Wrong question to ask. How often does it have to slow down to 5 mph or pull over and stop because it doesn't understand what's going on??? Or the computers freeze up?

I'm guessing that's pretty frequent. Otherwise, why don't we have reporters going on trips around the city and writing about the experience?

These pieces of crap are going to make bad traffic even worse. Hailing a ride from one of them will be worse than waiting on a bus. After a few riders, they will stink and be full of trash/vomit. Imagine one of these taxis after a few dindos take it for a ride.

dilligaff's picture

If I see one of these following me in Phx I'm going to slam on the brakes and hope it hits me.  That's my new retirement plan!

MagicHandPuppet's picture

Good thing they'll never be hackable/reprogrammable by a moozie.

Oh, wait....

Ristretto X4's picture

I predict that you will see drivers swerve in front of these "Johnnie cabs" followed by locking the brakes. This will result in law suits taking these cars off the road.

Let the fun begin!

booboo's picture

Death is a small plice to pay for FreeDumbs (or sumfin like that)

tmosley's picture

Your prediction is bad and you should feel bad.

You can pull shit like that with a human because there is no record and humans have bad response times. The AI can hit the brakes almost instantly.

CaptainObvious's picture

True, AI can hit the brakes almost instantly.  AI cannot defeat the laws of physics, however.  If there isn't enough time to stop a vehicle moving at high speed on a highway, the magic of AI will not slow down time or lengthen the road to prevent an accident.

tmosley's picture

The attacker's brakes only work so fast.

If you are talking about other things, then the AI will respond to those quicker than a human could too.

mkkby's picture

Computers never freeze and always do what you want instantly, right Einstein?

Can't wait until these white elephants hit the showroom floor. You want $200k for a van that's completely stuffed full of computer gear? Fuck that.

Agent P's picture

Why does a fully autonomous car need side and rearview mirrors?