Waymo’s autonomous vehicles are driving 25,000 miles every day

https://techcrunch.com/2018/07/20/waymos-autonomous-vehicles-are-driving-25000-miles-every-day/

by sahin-boydas • 8 years ago

153 178 8 years ago

Fricken2• 8 years ago

Waymo has tested in 25 cities, they aren't currently in 25 cities. Of the 600 Pacifica's they have, only ~160 are deployed. So to do 25,000 miles per day, that's an average of 156 miles per vehicle.

blacksmith_tb• 8 years ago

That does seem like a lot, I suppose they theoretically could be driving 24/7/365 (the wear and tear would be serious, you'd think).

The idea of moving people to public transit sounds good, but I wonder if there's enough spare capacity in most transit systems to serve lots of new riders...

vkou• 8 years ago

There is spare capacity in most transit systems to accommodate new riders, just not at the SLA that prospective new riders are accustomed to.

If your suburban bus stop is a 10 minute walk from your home, and the bus comes by once every half-hour, and then you have to transfer to an every 20-minute bus, that is sufficiently offset to make the transfer take time... You aren't going to have a good time on your commute.

nostrademons3• 8 years ago

That seems pretty reasonable - 8 hours/day at 20mph, 5 hours/day at 30mph, and 2.5 hours/day at 60mph. Given a mix of highway, rural, and city driving, 156 miles/vehicle sounds about right.

JoshuaEddy• 8 years ago

And only 6.5 MPH if they're running 24 hours/day

trhway2• 8 years ago

5B miles in simulation. I wonder do they source the situations for simulation from Youtube from crazy/fail/accident driving videos.

y4mi• 8 years ago

I can't imagine these poor video streams are useful inputs for the algorithm...

Maybe simulations are when the human is controlling the vehicle and the algorithm is only allowed to .. simulate what it's actions would be if it were in control?

toast0• 8 years ago

If they've been recording everything, they can run the previous inputs on new algorithms and check the results, and also simulate sensor failures and what not.

Running the system on recorded inputs to see how it compares with human output is good too.

Fricken2• 8 years ago

When they encounter interesting situations while testing on real roads, they then take that situation over to the simulator and train their algorithms on 1000s of subtle variations of the situation, which Waymo calls 'fuzzing'.

Simulating effectively is non-trivial. In spite of testing on a comparable scale to Waymo, Uber made very little progress in 2 years because they couldn't get their simulator to work right.

ceejayoz• 8 years ago

I can't imagine these poor video streams are useful inputs for the algorithm...

Store a rolling last X minutes of full sensor data into memory, then dump it to disk if the supervising driver hits the "that was a fuck-up" button.

repsilat• 8 years ago

They don't use the video data in the simulations, they use the processed "Where everything is" stream. That makes it more robust to sensor tech upgrades, but less to errors to do with bad sensing or sensor failure.

Computationally cheaper and simpler, no doubt, too.

kickopotomus• 8 years ago

The source the simulations from sections of road that their cars had trouble with.

https://www.theatlantic.com/technology/archive/2017/08/insid...

tim3334• 8 years ago

That would be entertaining to watch. Possibly also useful to check the cars don't freak in weird situations. They kind of had a real one in the Chandler crash https://www.youtube.com/watch?v=vxqBS2-4puw

dzdt• 8 years ago

Human drivers average about 1 fatality per 100 million vehicle miles traveled. [1] So we are still a long ways from being able to access safety in comparison to humans. [1] https://crashstats.nhtsa.dot.gov/Api/Public/ViewPublication/...

hcnews• 8 years ago

Number of fatalities per million miles driven is only one of the metrics. Hours spent per lifetime driving is arguably a more important metric. I can't wait for everything (cities, highways etc.) to be redesigned for self-driving cars but looks like its not going happen in my life time :(

pvarangot• 8 years ago

If you have to sell self driving cars as "more people will die but some people will have more free time" it's not going to be as easy as if they are unarguably safer for everyone.

ummonk• 8 years ago

I mean, it is already hard enough to redesign infrastructure for trains, trolleys, buses, etc...

ucaetano3• 8 years ago

I can't wait for [cities] to be redesigned for self-driving cars

Personally, I hope it never happens. The US is still suffering from when its cities were redesigned for cars, vs. public transit and pedestrians.

Self-driving cars can and will play a very important role in transportation, but cities should be designed around people, not cars.

Edit: I'm by no means criticizing your hope, and I do agree with you that cities should be redesigned to "accommodate" self driving cars, such as "buffer garages" (where autonomous cars are parked in a FIFO way, not in the way cars are parked today), auto-charging stations, smart curbs for pick-up/drop-off, etc.

gok• 8 years ago

Seems like you could estimate fatality rate from crash rate? From your source, about 1 in 200 crashes are fatal.

Skunkleton• 8 years ago

Yeah, but humans crash differently than robots. The fatality rate for human crashes does not apply to robot crashes.

bmcusick• 8 years ago

There's more to safety statistics than just the mortality rate. We should be able to look at frequency and severity of non-fatal accidents too. Those happen a lot more often.

For example, accidents that occur at 35 MPH or less are much less likely to result in a fatality or major injury, due to the amount of kinetic energy that a human body can safely dissipate. So if Google cars have even slightly better braking or speed control, you're going to see an improvement. Looking at the average speed at which accidents occur would be useful information.

YeGoblynQueenne7• 8 years ago

> There's more to safety statistics than just the mortality rate. We should be able to look at frequency and severity of non-fatal accidents too. Those happen a lot more often.

When the industry or the press discuss the safety of self-driving cars, they almost always do it in terms of fatality rates. Not general accident rates.

agumonkey5• 8 years ago

I wonder if there will be a psychological factor too. If SDV really manage to get better, during the transition, will the remaining human drivers improve to avoid feeling beaten by a machine ?

kuprel• 8 years ago

Also, some people are more reckless than others. A responsible driver might have a much lower chance of fatality than the average

jkaptur• 8 years ago

That sounds like common sense, but do you happen to know if it is actually true?

icc97• 8 years ago

Waymo have driven 5 billion in simulation. I think simulation is going to be the major change that automous vehicles can bring.

Assuming simulation is proven to be good enough for training cars, then having the ability to simulate crashes so that cars can respond in an optimal way is going to make a massive difference.

Animats3• 8 years ago

Waymo uses simulation to fuzz test their algorithms. When real driving records a troublesome situation, they load that data into a simulator and run it over and over with variations to see that the results are safe. Remember the problem with machine learning - most of the time it's right, and sometimes it's really, really wrong for no visible reason.

prolikewh0a• 8 years ago

People almost always tell me autonomous vehicles are much safer than Human drivers, but that has not yet been proven [1]. Luckily Tesla, Waymo, etc. are starting to put major amounts of mileage into the data.

https://orfe.princeton.edu/~alaink/SmartDrivingCars/Papers/R...

rickycook2• 8 years ago

well, that’s not really the right question; given enough time, they’re always going to be safer than human drivers

the question is at what point we decide that it’s “good enough” to be a life saving compromise

humans are really terrible drivers compared to sensors. the number of inputs, precision, and quick decision making are all much better for driving than humans

RestlessMind• 8 years ago

As an HNer, I find what Waymo (and other autonomous vehicle companies) are doing is simply fabulous. Technology improvement is a big impetus for human progress. Looking back over the past century, inventions like transistor, rockets, internet, satellites, nuclear power and aviation significantly improved our quality of life.

Autonomous vehicles belong to the same league. I am very happy that a non trivial amount of resources are devoted to them and it's not just a next SV fad like some ICO / Juicero / photo sharing app. Rooting for the success of this technology.

mrlatinos• 8 years ago

This isn't Reddit, no need to prefix an otherwise normal comment with internet credentials.

IshKebab4• 8 years ago

"As a mother..."

YeGoblynQueenne7• 8 years ago

Cars destroy the environment (with the infrastructure needed to accommodate them) and are one of the primary drivers of climate change. We need technology to make them obsolete, not to make more of them that can drive themselves (or not).

eboyjr• 8 years ago

Thanks for playing Devil's Advocate. This technology applies to mobile transportation in general and it doesn't necessarily have to effect greenhouse gas emissions. Additionally, I believe the infrastructure for producing cars applies to such a broad category in regards to driving climate change. The manufacturing industry needs to independently solve this problem.

nugget• 8 years ago

Won't autonomous driving eliminate something like 80 to 90 percent of private vehicle ownership, which is well on the way to obsolete?

Once people are less tied to private vehicle ownership and open to any effective form of transportation, then adoption of new technology could become easier.

jliptzin• 8 years ago

Humans destroy the environment, maybe we should get rid of them.

I don't see how greenhouse emissions is relevant to this conversation. Obviously, we'd prefer to all be driving around (human operated or otherwise) in vehicles powered 100% by clean energy, and certainly people are working on that, but that's independent of whether or not those vehicles are operated by humans.

nostrademons3• 8 years ago

If you want to make them obsolete you need a rival technology that possesses all of the attributes that make people choose cars over mass transit, notably convenience, scheduling & destination flexibility, privacy, and cargo capacity.

Failing that, self-driving cars as a service preserve all of these while also being significantly better for the environment. You don't need parking lots, because utilization is close to 100% and they go back to a depot when not required. It's easier to electrify the fleet, because the service knows exactly how far the trip will be and can bring the car back to a charging station while it has no passenger. You can share rides when multiple passengers have destinations that are close without them needing to go to the exact same place. You can have pods that dock with each other in motion (benefitting from better aerodynamics) but preserve privacy and split off to different destinations when needed.

boulos• 8 years ago

Fwiw, the Jaguar i-Pace that Waymo will be using in addition to the minivan is all electric. Getting people into electric vehicles gives us the option to recharge them via power generated by renewables. That's a great way to reduce the rate of climate change, even if it's a small step. (The minivan helps if it really gets people to drive with more people per ride)

tuxdev• 8 years ago

Autonomous cars could allow space currently devoted to idle car storage to be reused for denser development patterns, which would reduce the need for car-based transport in the first place.

njoro4• 8 years ago

While I hope the technology succeeds I don't think it is going to be as big of a change as people think. If you are upper middle class in a large city where labour is cheap, like in some Asian countries, driving is already close to free. And while that is somewhat nice from a quality of life perspective, it also means that rush hour traffic is horrible and sitting in a car isn't much more fun just because it is cheap. Basically a car is still a car.

airstrike4• 8 years ago

The idea is that with autonomous driving nobody needs to own a car and traffic jams are severely reduced because robots don’t drive like assholes (or at least by design shouldn’t...)

kevdragon6• 8 years ago

Reducing driving accident deaths (>37,000/year in the US alone) will be a pretty amazing change.

2bitencryption• 8 years ago

Uber went from zero to essentially killing off taxis, despite the fact that "a car is a car".

Primarily because they don't pay as much for their fleet as a taxi service does.

What's coming next is going to be that, except you don't have to pay your drivers anything because you don't have drivers!

(yes, there will be huge technical overhead of running this system and maintaining the cars/hardware, but over time this will become mass produced, and become cheap).

clomond• 8 years ago

However, I think you should consider how autonomous vehicles enable better utilization of existing cars on the road. If you own a car, what % of the time is it sitting in your driveway idle?

Rather consider how with autonomous driving, when paired with services like Uber/Lyft enable a car to be active 80-95% of the time - driving folks around. This means that your cost “of the trip” can be significantly less once you factor how his affects vehicular depreciation.

bodas• 8 years ago

Optimistically, self-driving vehicles may allow the west to finally discover the joys of share taxis, which would increase people/vehicle and (hopefully) decrease congestion.

VikingCoder• 8 years ago

"“Our experiments show that with as few as 5 percent of vehicles being automated and carefully controlled, we can eliminate stop-and-go waves caused by human driving behavior,” Daniel Work, assistant professor at the University of Illinois at Urbana-Champaign and a lead researcher in the study, said in a statement."

https://www.rdmag.com/news/2017/05/study-shows-self-driving-...

Jyaif• 8 years ago

Autonomous vehicles will make deliveries become almost free. This in turn will make food delivery explode, to the point of kitchen becoming an option in houses. In general, owning stuff won't be as important as before because you'll be able to simply borrow anything you need.

tbrock2• 8 years ago

Do they really belong to the same league? I’m sure this will be downvoted into oblivion because everyone on here is working on these projects but I don’t think self driving cars should be uttered in the same breath as any of the aforementioned technologies. It just doesn’t matter all that much.

Maybe I’m being dense but can soemone spell out the glorious moment we are all waiting for here after spending all this money?

- Trucking is cheaper because we fire the drivers?

- There are cheap Ubers taking people from ramen to sushi in SF but clogging the streets?

- There safer roads because there are no bad drivers?

Ugh, this is not a transportation revolution folks. Where are the personal transportation devices? The hypersonic planes? The hyper loops?

newfocogi• 8 years ago

Well optimized, high volume autonomous taxi services could be revolutionary: imagine if cars being parked went from being the norm to being an rarity. Instead of 90% of cars in cities being parked, there could be drastically higher usage rates. The average apartment complex wouldn’t need to be 50% parking lot; cities wouldn’t be rows of parallel parking. I can’t find the reference, but a surprisingly high amount of useable space in densely populated areas in the US is parking lots. It wouldn’t happen overnight, but if cars were optimizing for time spent moving people, a city could be much more space efficient.

keymone• 8 years ago

So if it prevents thousands of people from dying it somehow isn’t a big enough impact? Or tens of thousands from being injured? Or makes co2 emissions 10-30% lower? Or makes delivery of goods same amount more efficient? Are you overlooking all these gigantic benefits to social and economic life intentionally? I don’t get it...

michaelkeenan• 8 years ago

This is a good question, and doesn't deserve a snarky response.

This is the long-term good-case scenario (using USA-focused numbers, but you can extrapolate to the rest of the world):

-Car accidents will be reduced to nearly zero, saving about 34,000 lives and 250,000 injuries each year. The NHTSA estimates that this costs $871 billion per year[0].

-Because car accidents will be much less frequent, cars can be built lighter, saving the energy that's currently spent on armor. And in the scenario where people hail rides rather than own cars, the car that transports you will be the right size for you - i.e. a one-person car for most trips - which saves a lot of energy too. The energy savings help with intra-city air pollution, climate change, and decreased costs.

-Self-driving cars free up the 50 billion hours that Americans spend driving each year. At the average wage, that's worth about one trillion dollars.

-A lot of valuable land is used for parking (14% of Los Angeles, for example[1]). Hailed robot cars wouldn't need to park, and self-owned cars can drop their owners at their expensive destinations in a central business district, and drive a few minutes away to park in a garage where they pack together much more closely than humans can park them. The freed-up space can be used for housing or other valuable uses.

-Hopefully, self-driving cars massively reduce congestion. Due to their much-faster-than-human reaction time, they can drive very close behind each other at high speeds. Two single-person cars can fit side by side in a lane. Street-side parking space can be turned into more lanes. Reduced crashes result in fewer traffic jams and lane closures. (Personally, I worry that increased car usage might exceed these effects.)

Most of these figures come from Brad Templeton's webpages about robot cars[2].

[0] http://thehill.com/policy/transportation/207567-nhtsa-car-cr...

[1] https://www.citylab.com/transportation/2015/12/parking-los-a...

[2] https://www.templetons.com/brad/robocars/robot-cars.html

brudgers3• 8 years ago

With the possible exception of MAD geopolitics, I can't point to instances where nuclear power has significantly shapes my day to day experience in the US. Certainly much less than hydropower. Sure Voyager and other nuclear powered space craft have provided infotainment for many years, but terrestrial electricity is fungible for the end user and nuclear power plants have always correlated with more expensive watts when I've had utility service that owned them.

iknowstuff• 8 years ago

I feel like MAD is pretty significant by itself. Might not have been as peaceful otherwise.

cosmon0t• 8 years ago

A serious question here, and not trying to ruffle any SV'ers feathers, but why do you see autonomous vehicles to be as valuable as...say transistors, or satellites? In terms of long term usefulness for the 99% who care much more about turning up at work on time and/or transporting things efficiently from one place to another.

hgdsraj• 8 years ago

Autonomous vehicles could optimize routes and driving to get things to places faster, reduce stress while driving, and create safer roads. If the tech was widespread enough, there could be autonomous lanes that can drive at very fast speeds too.

jballanc• 8 years ago

My favorite explanation of why autonomous vehicles will be so game changing is this imagined future conversation:

[30 years in the future...]

My son: Wait...you mean to tell me that people used to operate multi-ton machines, with a minimum of training, in a range of different conditions, and relied solely on human reflexes and human attention span to avoid collisions?

Me: Yes

My son: And nobody was killed?!?

Me: Oh no! People died ALL THE TIME!

---

Look at any source you like, and it's as clear as day: the LEADING cause of death for individuals in their prime productive years (late teens through late 30s) is the automobile! If autonomous cars are even a fraction safer than human drivers, the impact it will have on demographics and population productivity will be MASSIVE.

droopyEyelids• 8 years ago

...the company is also working to apply its self-driving system to three other areas, including logistics (so trucking)...

This whole 'disenfranchised underclass' thing that America is struggling with will get considerably worse when automation turns all the truck drivers and their families out onto the street.

RobertoG• 8 years ago

When automation arrive to that level, that could happen, but it's far from unavoidable.

If automation happens is because it makes the economy more productive. If before happening the economy could sustain all the people, logic tell us that now could do it even more easily.

At the end, it's a political decision.

CydeWeys• 8 years ago

Most people used to be farmers, too, but prohibiting industrialization was not the right call.

Driving trucks is a dangerous occupation that's terrible for your health, and often takes you away from your family for long periods of time. In a few generations when it's long gone as an occupation, no one will miss it. Getting from now to then is unfortunately going to be a rough transition for some people, but stopping that transition entirely isn't the right call.

EthanHeilman• 8 years ago

Most people used to be farmers, too, but prohibiting industrialization was not the right call.

The industrial revolution caused enormous suffering for ex-farmers. Many of them lost a life of autonomy to be forced to move to cities and work in dangerous conditions for starvation wages. It caused significant social instability.

https://en.wikipedia.org/wiki/Industrial_Revolution#Standard...

IshKebab4• 8 years ago

And I think that is the place it will happen first. At least outside America where most roads are far too complicated for driverless cars.

Motorways, on the other hand are relatively simple. And we have service stations that driverless lorries could drive themselves to, to be picked up by human drivers for the final leg. And there's an obvious commercial need. And there's an obvious safety need (who hasn't seen a swerving lorry?)

tlrobinson2• 8 years ago

Waymo is supposedly planning on launching a ride-hailing service. Perhaps they can start giving rides to customers while continuing to test with a live driver (to take over if necessary). The revenue from the ride-hailing service could offset the cost of testing, and allow them to operate indefinitely until full autonomy is ready and legal.

bob_theslob6462• 8 years ago

Unfortunately the liability is too high for them to start taking passengers without thorough.

The idea is pretty clever. People paying to be guinea pigs.

Sure if people signed off, it would reduce some of the liability but that's still the golden question: who's responsible if something happens?

ghaff5• 8 years ago

IANAL but seems pretty obvious. If I'm just sitting in some company's product and it causes harm to me or someone else, it's 100% on the company. Of course, anyone can be sued for anything but the idea that I could be criminally negligent because an autonomous vehicle I'm sitting in (whether car or train) killed someone seems absurd.

jacksmith21006• 8 years ago

Waymo has insurance from Munich Re. Do not think liability problem.

https://www.insurancejournal.com/news/national/2017/12/20/47... Startup Trov, Munich Re to Insure Riders of Waymo's Self-Driving ...

tlrobinson2• 8 years ago

You can insure just about anything. At some point Waymo will be able to show the risk is low enough that insurance will be economical, if it’s not already.

bdod6• 8 years ago

ride sharing isn't exactly very profitable at the moment. I don't think the revenue would offset the additional engineering/product costs of just running the rideshare service, much less the cost of testing.

tim3334• 8 years ago

Waymo Early Riders Can Hail Actual Driverless Minivans Now http://fortune.com/2018/03/13/waymo-driverless-minivans-phoe...

vannevar• 8 years ago

The number of miles is not a sufficient metric; the question is, how much of the automotive phase space (weather, traffic conditions, terrain, road surface type, time of day, etc) has been explored? Running over the same mile a million times does not add anything to the self-driving model.

scarmig2• 8 years ago

Interestingly, going with that logic disengagements per mile will at some point correlate with the most advanced AV companies, not the least.

I guess the response would be to target some SLO budget and, once disengagements get too high, refocus on existing cases to get it down to an acceptable level.

agumonkey5• 8 years ago

I'm still waiting for readings on n-SDV interactions too. I can sense the emergent resonance failure coming.

YeGoblynQueenne7• 8 years ago

> Waymo has “driven” more than 5 billion miles in its simulation, according to the company.

Impressive. That's roughly equivalent to about 0 miles driven in real-world conditions.

bob_theslob6462• 8 years ago

I'm assuming sarcasm. What other way would you have them test besides current ways that they are doing now?

brudgers3• 8 years ago

To me, the very existence of a self driving vehicle is an amazing technical achievement against a really hard challenge. I suppose that explains why the math makes sense. 25,000 aggregate miles/day for a fleet of 600 vehicles is below 42 miles/day/vehicle. Each vehicle is providing a level of service below 2 miles per hour. Again, as a geek I think it's amazing but not convinced it is going to scale to the point that will justify those governors pushing public policy changes that impact the general public...at least for now.

lewis500• 8 years ago

1042 miles per hour is a fast car

oddmind• 8 years ago

Are Waymo using AI and humans to build a ginormous curated decision tree that can be audited or are they doing something more black-box / pure AI / something else?

I always imagined a big decision tree that could get diff’d after each AI training run and tuned by an army of humans.

ebikelaw• 8 years ago

Why do you assume there's any "AI" in it at all?

IshKebab4• 8 years ago

Because this would fairly obviously be impossible without ML techniques.

tim3334• 8 years ago

There's some info on the systems here https://techcircle.vccircle.com/2018/05/09/how-google-s-waym...

“While perception is the most mature area for deep learning, we also use deep nets for everything from prediction to planning to mapping and simulation.” ...Dmitri Dolgov, CTO and VP, engineering, Waymo

etc

Waymo’s autonomous vehicles are driving 25,000 miles every day

Related Stories