Waymo has a big lead in driverless cars—but here’s how they could lose it

Even if Waymo gets to market first, it'll face stiff competition from rivals.

Read the whole story

 

[Blarglefartz]

If you haven't experienced what it's like to be able to pay attention without having to make every decision, you honestly have no real understanding of what it's like. It's fantastic, and it's something I hope more and more car companies do.

I believe you, though I haven't been able to afford the joys of a Tesla (and probably never will just because I personally detest the likes of Elon Musk and everything he touches, but that's way off topic).

I regularly commute between my two homes which are 250 miles apart (divorce in progress) and I would absolutely love what you describe as a "level 3" car. As it stands, I take Amtrak. I expect the advent of this technology might reduce their ridership.

Thanks for relating your experience.

 

[melgross]

It’s a big mistake to think that anyone has a lead in this now. We won’t see a “real” autonomous car on sale for a good year yet, maybe two.

The first isn’t always the best (often the second or third is better). It doesn’t always garner good sales either. We won’t know who is in the lead for several years after these are on sale.

We don’t even know who is in the technical lead, though there is this assumption. Assumptions aren’t always right, and writers on the subject know no more than we do, because the real information is behind a wall of secrecy.

 

[unequivocal]

If you're a long-haul trucker and you're not looking to retire in the next 5 years, I'd get started on retraining now. Skilled trades are a growth sector right now. Plumbing, carpentry, steam-fitting, electricity, any of those are a good plan and you'll make more money to boot. If you don't move on this soon you're going to find yourself without any good prospects.

Why? It's the easiest job to automate, keep those self-driving trucks on the big highways and it's much simpler than city driving. Automated long-haul trucks are going to be the first automated vehicles we see.

3.5M angry, unemployed long haul truck drivers voting for president in 2020 and 2024 scares the shit out of me.

 

[Blarglefartz]

Level 3 cars absolutely do not let you fall asleep for 200 miles. Think adaptive cruise control and lane keeping on steroids: they assist you but you, the driver, are still required to pay attention and be ready to take over at any instant.

Level 4 does fully automated driving in some circumstances: perhaps freeway only will be one of those, although I'm betting it'll be geofenced to downtown urban areas.

4 and 5 are the only interesting levels. Current cars with level 2 are nice to drive and it'll be ok if some level 3 capabilities sneak in, but I'm still driving the car. It's just easier and less stressful when the car watches lanes, distance I'm following, etc.

Thanks for making those classifications clear.

I'll correct myself then and say I wouldn't be attracted to anything less than a level 4 car. What I'm looking for is a vehicle I can put on a freeway and forget. Watch a movie, read a book, or fall asleep. The same sort of experience I get on a bus, train or plane. Anything less than that has no real value for me.

EDIT: Spelling

 

[BasiliskPie]

Long haul Anything - cars or trucks - present more problems than seem easily negotiated by an autonomous vehicle.

I've driven Wyoming highways at night in fine weather, but where the roadway was obscured by a low fog of snow blowing across.

I've driven a freshly paved western Interstate where they hadn't striped the lanes yet and your only guidance was reflector poles (for plows?).

I've encountered accidents where troopers waving cone-lights directed drivers to merge and drive on the shoulder.

Many broad toll plazas can have approaches with unstriped areas where drivers are expected to choose between curving lanes ahead - where there's ambiguity as to which toll booth an opening lane leads.

On a minor highway, I've passed through after a windstorm created a slalom course of tree limbs.

Intra-city issues abound with areas of roadwork, but at least support-people could be on hand,. Outside populous areas, who steps in to solve the unexpected problems? My trips out west have left me without cellular on many occasions, so are autonomous vehicles to have video-bandwidth satellite communications?

On the bright side, I'm looking forward to the adventures of long drive passengers: surely someone in Ohio will load their skis and program their destination as Winter Park, CO, only to wake up well down the road to Winter Park, FL.

 

[duedadew]

really? youd trust an uber branded self driving car with your life?

That, and the backup driver, yeah. But then I trust other drivers not to kill me every time I get behind the wheel, and that's a hell of a lot worse.

uber isnt ahead because its a ponzi scam 80% of the rides dont cover drivers cost in 90% of markets its literally 80% slavery & if you didnt tip least $5 a ride and didn't go more than 7-10 miles your driver lost money.

if you don't get the same driver regularly theyre unmatching from you ; )

So what you're doing is taking my comment about a very specific thing that Uber does, and using as a launching point for every petty grievance you have with the company, completely ignoring what the discussion was about?

Yeah, great. I get it, you don't like Uber. Cool story.

Regardless of your personal feelings, Uber has had an actual program dropping customers off in Pittsburgh for over a year. Waymo is just now launching their own. While I'm prepared to accept that Waymo's tech is better, I think the article, in claiming that Waymo is "launching a self-driving car service before anyone else", isn't factually accurate. Uber did the same a year ago.

Whether you hate them or love them, they're one of the groups pushing the technology ahead, and it seems odd to claim that Google is the first to launch. Maybe the technology is so much better. But it doesn't sound like Ars has used the Uber service, so I'm just interested in why they don't seem to be considering it.

There may be a perfectly valid reason for that. Your spleen-venting about Uber, however, isn't that.

i love 20% of uber or the concept of it, the people running it are pure evil their actions shows that...


....Sufficiently advanced incompetence is indistinguishable from malice....

youre telling me a human adult actually thought a car only costs/depreciates $500 in a year of constant driving? they had to shutter theyre exchange predatory lease program because they miscalculated by 18.5 times????? wheres that in the new hire pamphlet why arent they shuttering the pool x tiers? theyll stop losing money themselves but theyll let drivers who use their own vehicles to take the loss? thats not a legit job


i hate slavery which 80% of uber Literally and technically IS. eventually a real journalist will get it out there, its not my profession....drivers lose money on every x or pool ride less than 7-10 miles depending on vehicle type to the point 96% of their work force fail and only 4% succeed more like a pyramid shape then a legit business...

just posting facts no one seems to care about since real human lives are being exploited nevermind the law suits, settlements for fraud, bait & switch, among other things cheap rides are cool i get it just ignore the fact these people are flushing money & people down the toilet at 41% of actual costs which is predatory if not evil well that is in a fair free market but oh well lifes hard but fair keep marching on people are now loss leaders i suppose

charging considerably below costs, paying a 1971 minimum fare, paying 1965-1985 wages in 2017....isnt pushing tech along it what maybe 1000 billionaires on the planet can do & your best interests are not their top priority there is a reason poor people don't have private drivers/chauffeurs, cabs arent meant to be taken daily on the vast majority of the planet

i could get lots if people to use my crack/pussy/meth sharing app too, with 13+ billion to burn and 6.5+ more million to lose daily its not rocket science it is disruptive though remember your driver got $2 in 2017 for what they did and he has a 96% chance at failure and their reward at the end of the year is a worthless car they couldn't afford to pass 2nd inspection but self driving cars in 2019 hahaha

 

[Double-T]

Currently, completely autonomous cars at not legal anywhere. A driver is always required.

This is incorrect. Completely autonomous vehicles are permitted in Florida.

http://www.politifact.com/florida/state ... florida-s/

 

[drdiatom]

No mention of Google's customer support?

I think many of us have experienced their black hole on the occasion we needed help or assistance. I can't even get their flagship hardware delivered to my house or products returned/replaced without being stuck in a customer support hell hole for the last week, and looking at any official Google support and help forum you'll see a bunch of others with similar issues.

Google doesn't care about their customers when they get stuck. Probably because 99% of them don't experience issues, but boy does it suck to be one of the unlucky few.

Good luck getting help from Google when you need it.

 

[Foiler]

I think the problems of the arms length relationship is way overstated, upfitters have been moving thousands and thousands of customized production units for forever with little or no cooperation from the OEM. I don't think anybody is realistically looking at millions of units at this stage in the game, if it does look like Waymo's tech is up to that level of quality then I'm sure they'll be able to find a partner or somebody willing to buy them as an Alphabet spinoff.

Not to mention the tech itself has to be mass produced.

ultimately until the end of the 2020's I don't expect self driving cars to be more an expensive taxi style service which means a million cars at the high end.

To get them into consumer hands requires the tech itself to drop Dollar wise. currently a self driving car is a $150k investment. With the car being only $35k of that(price of a pacifica).

That will be the big barrier to adoption.

Why would there be a need to get them into consumer hands?

If there's cheap selfdriving cars driving around, I wouldn't to purchase a car. Why pay for a car, when I can simply share the costs with all the users. Not having to pay insurance is a boon as well.

You do realize you are still paying for insurance, right? It is rolled into the cost whether it's listed separately or not. In fact, we all share the same insurance pool. I will be forced to supplement the insurance on these self-wrecking vehicles whether I like to or not. Queue the multi-billion dollar class action lawsuit as people get maimed/killed from them.

 

[Blarglefartz]

there's no technical solution for the fact that lots of people live in low-density areas but work in high-density areas, that travel demand is peaked (i.e. rush hour exists), etc.

Exactly, and that of course is the reason suburban and rural areas have poor taxi service support. The same would be true for autonomous vehicle fleets; it just doesn't work economically without a high population density. But it would work very well in those high density areas, and let's face it, that's where most people live and work. It's also a demographic that typically doesn't have space for a personal vehicle. In most areas of San Francisco for example, a garage is very rare and parking on the street is inconvenient and usually expensive (parking tickets can run a person over $1000/year easily).

I still believe very strongly there's a market for what's been described to me as a "level 4" autonomous vehicle that can take over once I'm on a freeway. I'd buy that.

 

[river-wind]

I'd love to see a comparison between this and Uber's self-driving cars. As other commenters have noted, this is a potential revolution of the need for car ownership in the first place. I live in a big city, and the only time I drive is to leave it. But I order Uber/Lyft (whichever is cheaper) a couple times a week. The car-sharing aspect of it is going to change ownership habits a lot, I think. This is where Uber has a pretty good advantage.

Well that, and the fact that, you know...they already have real cars, driving by themselves, on real roads, serving real customers. For all the talk of how Waymo has a "sizable lead" in the technology, and how others are coming out with the service in the future...Uber already has a "self-driving + saftey-driver" system that you can actually experience. Waymo's tech may be better, but it's hard to say they're ahead of everyone, when Uber is actually using theirs, right now.

Maybe I'm just missing some special sauce in what Waymo is offering, but...why isn't what Uber has been doing considered "ahead" in this particular race? Is the Waymo tech that much better?

Edit: Uber's self-driving car pilot for customers actually started over a year ago.

I need to do more reporting on this, but I've talked to a number of industry insiders and none of them see Uber as a leader. Their cars were running red lights in San Francisco less than a year ago (http://fortune.com/2017/02/26/uber-self ... ed-lights/) and I don't imagine the chaos of the last year has given them room to catch up. I wouldn't count them out, but I highly doubt they're in the top tier at this point. A research firm called Navigant did a ranking earlier this year and rated them near the back of the pack. (https://www.navigant.com/insights/energ ... ng-systems)

I wouldn't call myself an industry insider, but from talking with people who are, my impression is the same. From being cavalier with public road testing before things are ready, to not being up front with the press about how often the human driver has to take over, to breaking laws regularly while on the road, Uber is the example of how not to get regulatory buy-in to self-driving cars.

They've fairly recently lost some of their top level engineers, and after their partnership with CMU died, their claimed position as the new leader disappeared.

The last demo from them I saw showed that the cars could not handle road construction or other obstructions which would require a complete route re-plan, something Waymo was doing 5 years ago under the Google Car moniker. In those cases, the person doing the demo took over control of the car to navigate the problem, not mentioning that they had to take over or why. It was misleading at best.

 

[Snarky Robot]

I need to do more reporting on this, but I've talked to a number of industry insiders and none of them see Uber as a leader. Their cars were running red lights in San Francisco less than a year ago (http://fortune.com/2017/02/26/uber-self ... ed-lights/) and I don't imagine the chaos of the last year has given them room to catch up. I wouldn't count them out, but I highly doubt they're in the top tier at this point. A research firm called Navigant did a ranking earlier this year and rated them near the back of the pack. (https://www.navigant.com/insights/energ ... ng-systems)

Thanks, I appreciate the follow-up. I know CMU was still reeling from the loss of so many of their employees, and I've seen the construction of the fake town, but I wasn't sure anyone had done a hands-on.

Look forward to hearing more on your experiences!

 

[Snarky Robot]

youre telling me a human adult actually thought a car only costs/depreciates $500 in a year of constant driving?

No, I'm telling you that none of your incoherent rants have anything to do with what I'm talking about. I'm going to add, based on this latest screed, that a properly trained monkey acting as a mime would do a better job of conveying your messages in a succinct and logical manner.

As I said, I get it--you don't like certain things. Cool story, bro.

 

[Blarglefartz]

As I say in the article, Arizona's governor seems to believe there are no regulatory barriers to launching a driverless car service in Arizona. And the feds are bending over backwards to allow these things on the roads. It's certainly possible that once a company tries to actually launch a service they'll encounter regulatory obstacles, but right now it doesn't look like it'll be a significant problem.

Thanks very much for following this story Tim, you've written an insightful and informative article, something I'd like to see more of at Ars.

 

[karoc]

there's no technical solution for the fact that lots of people live in low-density areas but work in high-density areas, that travel demand is peaked (i.e. rush hour exists), etc.

Exactly, and that of course is the reason suburban and rural areas have poor taxi service support. The same would be true for autonomous vehicle fleets; it just doesn't work economically without a high population density. But it would work very well in those high density areas, and let's face it, that's where most people live and work. It's also a demographic that typically doesn't have space for a personal vehicle. In most areas of San Francisco for example, a garage is very rare and parking on the street is inconvenient and usually expensive (parking tickets can run a person over $1000/year easily).

I still believe very strongly there's a market for what's been described to me as a "level 4" autonomous vehicle that can take over once I'm on a freeway. I'd buy that.

I agree with everything you say. Just wanted to clarify that I'm not actually sure, at least in the US, whether the places where "most" people live and work are of sufficient or insufficient density to support circulating driverless fleets. My intuition is that a ton of people live in low-density suburbs that are often considered part of a the city's "metro region," but that these will not be suitable for such fleets given the peaked demand structure and the higher marginal cost to serve each customer. I'd be interested in good data on such things.

And, of course, it will also depend on the costs of operating the service in different geographies, which may not yet be entirely known.

 

[Rene Gollent]

Only lightly related to the article, but kudos to Aurich on that headline image--and the graphics work in general!

It could only have been better if the background face was SHODAN. At least that's what my first thought was, but it's not quite the same.

 

[Blarglefartz]

Just wanted to clarify that I'm not actually sure, at least in the US, whether the places where "most" people live and work are of sufficient or insufficient density to support circulating driverless fleets.

Me either really, I'm just going by voting records as a proxy for population density; it seems to me areas like New York, San Francisco, Boston, London, Paris, Madrid and Seattle are high enough density to support taxis. Not sure about LA.

I guess I should have said "areas with convenient existing taxi service" rather than just "urban". I know it's hard to get taxis in places like San Mateo CA, which is only about 20 miles from SF. I can't see self driving cars (as a corporate operated fleet) replacing personal cars anywhere but urban cores.

 

[SlimSam]

...Aging baby boomers behind the wheel scare the crap out of me. And they're the ones who can actually afford to buy the cars with these advanced systems.

Might want to recalibrate your fears. Here's a quote from the first link I checked:
"The risk of motor vehicle crashes is higher among 16-19-year-olds than among any other age group. In fact, per mile driven, teen drivers ages 16 to 19 are nearly three times more likely than drivers aged 20 and older to be in a fatal crash."

Also,
"Driver error is most often the result of “distracted driving.” By far the most common cause of driver distraction is the use of cell phones. According to the National Safety Council (NSC), 25% of all automobile crashes are related to cell phone use, whether by talking on the phone or texting." - I think this is a smaller risk with baby boomers.

 

[traumadog]

It's interesting to me that the company mentioned here as the potential biggest threat to Waymo is GM, and not Tesla.

Kinda drives home the point that it's not the same company that went belly-up in 2009.

 

[ggalt]

What has always puzzled me is, why do all of these companies focus on self-driving **cars** instead of self-driving trucks. It seems to me that getting a truck to safely drive the highways is a significantly less challenging problem than avoiding the chaos of city driving.

Self-driving trucks solve a number of other issues too. You can get lots of miles of experience quickly, while not attracting too much attention (who ever sees the driver of a truck on the highway, so no one will freak out by seeing a car without a driver). You can greatly improve the throughput of a trucking company by having trucks operate 24x7. You can avoid the issue of sleep-deprived drivers on the highways. Etc.

 

[river-wind]

ggalt: many are. I see much less coverage of automated trucking, and I assume that's because more potential readers drive cars than drive trucks. If you look at publications dedicated to long-haul trucking, there are a ton of articles on the industry transitioning to autonomous driving.

https://www.trucks.com/category/tech/au ... -vehicles/

There are even trucks out today which can do a fair amount of the highway driving, or can be paired with a human-driven lead truck.

https://newatlas.com/daimler-truck-plat ... tes/51493/
https://www.technologyreview.com/s/6034 ... ng-trucks/

But that doesn't mean the transition from a human-driven fleet to an automated one will be smooth, or even guaranteed:
https://www.wired.com/story/trucks-robocar-senate-war/

 

[duedadew]

youre telling me a human adult actually thought a car only costs/depreciates $500 in a year of constant driving?

No, I'm telling you that none of your incoherent rants have anything to do with what I'm talking about. I'm going to add, based on this latest screed, that a properly trained monkey acting as a mime would do a better job of conveying your messages in a succinct and logical manner.

As I said, I get it--you don't like certain things. Cool story, bro.

ad hominem is ad hominem doesnt negate any facts that uber is 80% slavery.

i get it slave labor cool when its an app & since its in the app store must be legal who cares i get it a chauffeur at 41% of costs and fucking earth in the process who cares

i dont like evil or slavery, you do or dont really care world keeps spinning i dont try to decipher tone from text just posting real news not fantasy Jetsons paid advertising posing as a story that matters when the headlines should read how is uber legal when its 80% slavery

 

[co-lee]

youre telling me a human adult actually thought a car only costs/depreciates $500 in a year of constant driving?

No, I'm telling you that none of your incoherent rants have anything to do with what I'm talking about. I'm going to add, based on this latest screed, that a properly trained monkey acting as a mime would do a better job of conveying your messages in a succinct and logical manner.

As I said, I get it--you don't like certain things. Cool story, bro.

drivel

I try to keep it from getting personal, but gawd dude ...
I won't shed many tears when your job driving for uber is automated away.

 

[Snarky Robot]

blather

Yep, you have opinions that don't relate to anything I say.

Cool story.

 

[duedadew]

murica ef yeah

uber IS 80% slavery

carry on

 

[alxx]

I don't see the tech as the problem with self driving cars but the regulations.
Every country and every state has different regulations and insurance requirements.

All the self driving carsmentioned recently are targeted at the first world or on road.

Haven't seen much recent mention of self driving trucks for mining.

Rio Tinto (and others ) spent and spends a lot for its robotic/remotely operated mining sites and trains.


https://www.technologyreview.com/s/6031 ... th-robots/

http://www.afr.com/business/mining/iron ... 419-goa4l0

 

[Glindon]

If you're a long-haul trucker and you're not looking to retire in the next 5 years, I'd get started on retraining now. Skilled trades are a growth sector right now. Plumbing, carpentry, steam-fitting, electricity, any of those are a good plan and you'll make more money to boot. If you don't move on this soon you're going to find yourself without any good prospects.

Why? It's the easiest job to automate, keep those self-driving trucks on the big highways and it's much simpler than city driving. Automated long-haul trucks are going to be the first automated vehicles we see.

You’re still going to need a “driver” for semi-trucks. The second there’s no drivers in them is the second they get robbed out in the middle of nowhere. And since there’s no driver inside it’s a victimless crime to boot. Just box the semi in with a few cars and slow all the cars to a stop, forcing the truck to stop. Then it’s just a matter of getting inside and taking everything. Experienced thieves will be able to unload whole trucks in a matter of minutes.

 

[BilTheGalacticHero]

The question is, does Ron think the vans are premium enough? Should be enough metal and glass right?

 

[Project_2501]

If you're a long-haul trucker and you're not looking to retire in the next 5 years, I'd get started on retraining now. Skilled trades are a growth sector right now. Plumbing, carpentry, steam-fitting, electricity, any of those are a good plan and you'll make more money to boot. If you don't move on this soon you're going to find yourself without any good prospects.

Why? It's the easiest job to automate, keep those self-driving trucks on the big highways and it's much simpler than city driving. Automated long-haul trucks are going to be the first automated vehicles we see.

Whilst I don't disagree in general, there are additional issues to be considered with long haul trucks.

Long hauling requires refuelling. How easily is that automated? Will it require specialised pit-stop areas?
Long hauling may require boat/train/air transportation for the cargo. That will require a system to navigate the port areas properly and some way to let controllers at those ports tell the trucks where to go just as they tell people.
People at the dock may need to get in and drive the truck for some reason (position under crane?), that will require some sort of secure way to let authorized users in. Presumably you're not just going to post loads of key copies to places, lest people copy them and hijack the trucks en route.
If your route crosses borders it needs to co-operate with customs and somehow let them into the cargo area (or they just cut and then not reseal the tamper evident cable). So what does the truck do?

These are all solvable problems, I'm just making the point that while long-haul trucking may be easier from a automated driving point of view, it offers other challenges city driving doesn't. Waymo can just have their cars return to the depot when their fuel is low and have a human refill it, long haul trucks can't and need an en-route refuelling process etc.

Long haul truckers generally go from loading dock to loading dock.

they generally use a short run local truck driver to go from port to distribution point.

As a truck driver, this has been my experience.

 

[Hinton]

I think the problems of the arms length relationship is way overstated, upfitters have been moving thousands and thousands of customized production units for forever with little or no cooperation from the OEM. I don't think anybody is realistically looking at millions of units at this stage in the game, if it does look like Waymo's tech is up to that level of quality then I'm sure they'll be able to find a partner or somebody willing to buy them as an Alphabet spinoff.

Not to mention the tech itself has to be mass produced.

ultimately until the end of the 2020's I don't expect self driving cars to be more an expensive taxi style service which means a million cars at the high end.

To get them into consumer hands requires the tech itself to drop Dollar wise. currently a self driving car is a $150k investment. With the car being only $35k of that(price of a pacifica).

That will be the big barrier to adoption.

Why would there be a need to get them into consumer hands?

If there's cheap selfdriving cars driving around, I wouldn't to purchase a car. Why pay for a car, when I can simply share the costs with all the users. Not having to pay insurance is a boon as well.

Because, unless you live in a truly dense urban core that supports ubiquitous taxi service, waiting for a car to show up every morning to take you to work will be annoying as hell. It's not like it's going to be profitable for these companies to keep fleets of cars roaming around every low-density cul-de-sac in America (apologies if you live in Amsterdam or whatever). And god forbid you go in an hour late, after the vast majority of the fleet has taken people into downtown and is now idling waiting for them to head back.

Or what if you need to drive to someplace out in the countryside for the day? You'll be paying a premium to the company for the car to wait for you all day instead of returning to an economically productive (for the company) part of the region, or else waiting a long time for a car to get back out to you when you want it. The level of control over your own schedule will be a lot lower than if you owned a vehicle.

And of course with ownership comes an ad-free, other people's crap-free, "premium" experience.

It's possible people's expectations may change over time. But there's no technical solution for the fact that lots of people live in low-density areas but work in high-density areas, that travel demand is peaked (i.e. rush hour exists), etc. Owning will provide a far superior experience in many cases.

I live in something like Amsterdam.

And your points are valid.

Also, unlike the Ars authour, sorry I forgot his name, I found the discussion about whether X American company will make the driverless car work. The assumption that this will happen in USA, or will be made by an American company seems naíve to me.



But, the central point I made "why do you need to get this into consumers hands [to make it viable]"?

That's like saying, "the internet will never catch on, because some people live in rural areas". You know that is not true, well, obviously for the internet, because companies gives a shit, the same with the driverless car.


You're right, there will be times where there's a lack of cars, perhaps. Public transportation, paying a surplus for those times?

/Is it normal to make an internet analogy about cars?

 

[Picklesworth]

I still have trouble seeing why people take Uber seriously here. A good chunk of their user base is people who can't or don't drive, and possibly don't have licenses. It's interesting that some places are okay with this, but whenever self-driving cars start being a thing, there is no way regulators and the public in general (in enough numbers for this to be worthwhile) will allow those cars to drive on their own without a licensed driver ready to take control. Unless Uber's plan is to flaunt the law for (another) ten years, they're going to have a lot of trouble.

The people who are going to succeed here at scale are the established car sharing companies, and whoever provides them with equipment. They have cars, they have a system for maintaining those cars, they have dedicated parking spots, and all of their users are able and willing to drive - but I'm sure would be perfectly happy paying less for a self-driving car. They're in a great position to gradually roll in self-driving cars and the various wins from that technology, without needing a magic spell and billions of dollars in between.

 

[mjeffer]

I think the problems of the arms length relationship is way overstated, upfitters have been moving thousands and thousands of customized production units for forever with little or no cooperation from the OEM. I don't think anybody is realistically looking at millions of units at this stage in the game, if it does look like Waymo's tech is up to that level of quality then I'm sure they'll be able to find a partner or somebody willing to buy them as an Alphabet spinoff.

Not to mention the tech itself has to be mass produced.

ultimately until the end of the 2020's I don't expect self driving cars to be more an expensive taxi style service which means a million cars at the high end.

To get them into consumer hands requires the tech itself to drop Dollar wise. currently a self driving car is a $150k investment. With the car being only $35k of that(price of a pacifica).

That will be the big barrier to adoption.

When you're talking about driverless taxis, $150k is CHEAP if it makes your labor cost to operate that vehicle to $0.

 

[Netguru]

I think the problems of the arms length relationship is way overstated, upfitters have been moving thousands and thousands of customized production units for forever with little or no cooperation from the OEM. I don't think anybody is realistically looking at millions of units at this stage in the game, if it does look like Waymo's tech is up to that level of quality then I'm sure they'll be able to find a partner or somebody willing to buy them as an Alphabet spinoff.

Not to mention the tech itself has to be mass produced.

ultimately until the end of the 2020's I don't expect self driving cars to be more an expensive taxi style service which means a million cars at the high end.

To get them into consumer hands requires the tech itself to drop Dollar wise. currently a self driving car is a $150k investment. With the car being only $35k of that(price of a pacifica).

That will be the big barrier to adoption.

Why would there be a need to get them into consumer hands?

If there's cheap selfdriving cars driving around, I wouldn't to purchase a car. Why pay for a car, when I can simply share the costs with all the users. Not having to pay insurance is a boon as well.

You will still have those cost. Just as another line item fee..... Not to mention what happens when yo need a car NOW? No thx, I will want to own my car, not pay rent...

 

[MalikDama]

Why is SHODAN doing the driving, I really think this is a poor idea...

 

[BloodNinja]

If you're a long-haul trucker and you're not looking to retire in the next 5 years, I'd get started on retraining now. Skilled trades are a growth sector right now. Plumbing, carpentry, steam-fitting, electricity, any of those are a good plan and you'll make more money to boot. If you don't move on this soon you're going to find yourself without any good prospects.

Why? It's the easiest job to automate, keep those self-driving trucks on the big highways and it's much simpler than city driving. Automated long-haul trucks are going to be the first automated vehicles we see.

Whilst I don't disagree in general, there are additional issues to be considered with long haul trucks.

Long hauling requires refuelling. How easily is that automated? Will it require specialised pit-stop areas?
Long hauling may require boat/train/air transportation for the cargo. That will require a system to navigate the port areas properly and some way to let controllers at those ports tell the trucks where to go just as they tell people.
People at the dock may need to get in and drive the truck for some reason (position under crane?), that will require some sort of secure way to let authorized users in. Presumably you're not just going to post loads of key copies to places, lest people copy them and hijack the trucks en route.
If your route crosses borders it needs to co-operate with customs and somehow let them into the cargo area (or they just cut and then not reseal the tamper evident cable). So what does the truck do?

These are all solvable problems, I'm just making the point that while long-haul trucking may be easier from a automated driving point of view, it offers other challenges city driving doesn't. Waymo can just have their cars return to the depot when their fuel is low and have a human refill it, long haul trucks can't and need an en-route refuelling process etc.

Easy, you keep the port side work manual,.then kick the driver off and let the truck drive itself.

Refueling is easy as truck stops are all attended. What worries me is the human truck drivers at the same stops who may be tempted to sabotage the non human competition.

 

[BloodNinja]

I honestly wish companies would first launch level 3 autonomy vehicles first. I.e., I want a car I can be driving manually to be able to enter an automatic driving mode that just keeps me in my lane, doesn't hit anything, stops at red lights, etc. I don't even care if it doesn't know how to change lanes or turn onto different roads. Just let me get in a lane, press a button, and then zone out until the next 10 miles of my commute down the freeway are done.

The issue with that is you'll inevitably zone out for just 30 seconds too long and crash into somebody. Full independent driving is better than partial attention.

 

[exAstronomer]

The personnel issues at Telsa are nothing to sneeze at. However, I wouldn't count them out yet, for one reason: data

Google's early success in search came not just because they were bright, or even because their specific algorithm (PageRank) was great, but because they realized that everyone writing a web page was at the same time creating public data on the structure and quality of the web. They then parlayed that lead in search into an operation which gathered more data than anyone else in a succession of industries: search, web advertising, mobile advertising, voice search, etc.

Their success also allowed them to hire the world's experts in machine learning (including established ones like Geoffrey Hinton, and young up-and-coming ones), so they could be the best at taking advantage of huge quantities of data. But the data scientists without the most and best data wouldn't have achieved the same success.

So, what does Google have in the way of data on actual driving in realistic conditions?

3 million miles (https://waymo.com/ontheroad/), increasing at about 1 million car-miles / six months.

What about Tesla? Every single car that Tesla sells has the instrumentation to gather data on real driving conditions. Some of these data are only gathered when the autopilot features are enabled; not just when you are actively using them - see Tesla's privacy policy - https://www.tesla.com/about/legal). And Tesla's policy of rolling out software updates gives owners the incentive to connect their cars to Tesla periodically.

Tesla apparently sold 50,000 Model S cars in 2016 (http://www.hybridcars.com/tesla-model-s ... -in-a-row/) with total sales to that point exceeding 150,000. Tesla delivered over 45,000 cars in the first half of 2017 (https://www.recode.net/2017/7/3/1591673 ... arter-2017). Suppose their owners drive 1000 miles a year (at least the right ballpark - e.g. https://www.fhwa.dot.gov/ohim/onh00/bar8.htm). 200,000 cars * 1000 miles per year = 200 million car-miles / year, or 100 times what Google has. Even if 25% of owners decide to opt out, they'd still have 75 times what Google has.

Finally, take a close look at the parts of the Tesla privacy policy (https://www.tesla.com/about/legal) which mention data from accidents, and notice that those parts are not restricted to accidents while autopilot is in control or even enabled. That means Tesla can gather data on accidents which are the fault of the driver (or other drivers), not just the handful of cases where an accident is blamed (correctly or not) on autopilot, so they undoubtedly are accumulating lots of negative examples they can use to train autopilot what to avoid.

Google's data, by contrast, are gathered either from self-driving cars or cars driven by their employees, so either way they'd be seen as responsible for any accidents. That means they have to make sure that those cars operate only in safe conditions, severely limiting their ability to gather data on dangerous situations.

GM and Nissan sell more electric cars, but how many of them are luxury vehicles with the sensing equipment to gather that data, and how many of them have the technology to connect back to the manufacturer or the data use agreement to allow such data gathering. Unless other luxury car manufacturers can match Telsa's volume of instrumented, connected cars, no one else has that volume of data collected under real driving conditions including accidents.

Even if Tesla's entire team of engineers working on self-driving cars were to quit, leaving them unable to use all that data themselves, the data alone could be licensed to other players in the self-driving game for a significant sum of money; probably enough to license back one of those companies' self-driving technology and then some.
In a sense, Tesla isn't selling electric cars; they're running a huge data collection operation, and partially subsidizing it with the proceeds on the sale of the collection devices.

PS I have no financial stake in Tesla, and don't know of anyone who does. I've no training in finance, so while I suspect Tesla's data would be worth a lot, should they decide to license it rather than keeping it as a competitive advantage, but I don't know how its value would compare to the current TSLA stock valuation.

 

[got2bereal]

Automated Passenger cars is not the future for urban cities. We need less cars and better mass transit infrastructure upgrades.

Have you ever been to a big city? How's sitting in the car waiting for the red light and how long have you waited just to find a parking spot?

With Uber and Lyft it created more congestion for big cities with limited roads to handle the volume.

They should be reducing dependence on passenger cars and require only buses and high passenger car pooling options.

I like to see trams and trolley cars have automated or remote controls but right now it's still far too dangerous. A skateboarder could easily get run over if sensors don't work properly.

 

[Overlord_Dave]

...snip...

I've driven Wyoming highways at night in fine weather, but where the roadway was obscured by a low fog of snow blowing across.

I've driven a freshly paved western Interstate where they hadn't striped the lanes yet and your only guidance was reflector poles (for plows?).

I've encountered accidents where troopers waving cone-lights directed drivers to merge and drive on the shoulder.

...snip...

Ever watched C-beams glitter in the dark near the Tannhäuser Gate..?

 

[mmnw]

Currently, completely autonomous cars at not legal anywhere. A driver is always required.

This is incorrect. Completely autonomous vehicles are permitted in Florida.

http://www.politifact.com/florida/state ... florida-s/

Thank you for the informatin. I was unaware of this. Although, it seems a bit reckless by the state to loosen the insurance requirements.
That said, it also shows how fractured the legal landscape is.