Ninety5rpm

If even people on this thread are conflating simple mindless driver-assist technology with AI autonomous technology, it's no wonder the public is so dubious about it all.

northernlights

The Tesla in autopilot mode is considered a level 2 self-driving vehicle by the US government autonomous ranking system. The keyword words being self-driving and autonomous.

If the US government allows a commercial self-driving vehicle to be on public streets that many others, including Waymo, believe is inherently dangerous that is on them.

Leisesturm

Waymo isn't exactly a neutral party in this. I GET that they would have an issue with a Level 2 self-driving vehicle when their mission is to sell as many Level 5 systems as they can. When they can. And that isn't now. In the meantime though they will throw as much shade on 'lesser' systems as they can. But it simply isn't possible to win with you people. Level 5 systems are bad, period; Level 2 systems are bad because you still have to pay attention. Is that about right?

Maelochs

I think the issues might be two-fold.

First, human drivers just suck. No explanation or further discussion needed.

Second, as we see here---people cannot tell a car with a lane-change and emergency stop warning system from an autonomous robot. So, when a Driver is a Tesla screws up (not the auto-pilot system, which is NOT and auto-drive system but a name deliberately mis-chosen as a sales feature---the driver screwed up) then people jump all over all AI cars.

We can say "But, it is a Level-Two Autonomous Control device" or whatever ... but the fact is, a Driver-controlled car with certain Driver-aids was Driven by a Driver into a fire truck.

Waymo doesn't want to be tarred by the brush with which it hasd just been tarred ... and doesn't want stupid drivers crashing into its cars.

Think it through----has there Ever been a report of one of the Google or Uber or Serious AI cars just flat driving into a stationary object? No, because that is the First thing they are designed not to do.

A freaking Roomba can manage that.

This was a car with a lot of driver aids which the marketers decided to call "Autopilot" and which obviously does Not drive the car.

Anyone who can't grasp this, PM me ... I can get you Great deals on stairways to heaven, magic beans, a golden-egg laying geese.

I-Like-To-Bike

You are right, that you can't "win" with bogus arguments. You are wrong about what is "bad".

"Bad" is assuming that practical and useful level 5 systems are here now and ready to be produced, fielded, and used by the public in any meaningful way, or making pretend that predictions/assumptions of the imminent and/or sure-fire successful introduction of a practical and profitable Level 5 AV system (based on wishful thinking, PR hype, and/or references to smartphones et al.) are the same as facts.

Maelochs

Something between you and an un-named poster's unbridled negativity ...

level 2 systems are great as driver aids, but with drivers being human, statistically some number of them can overlook ten thousand warnings that the car doesn't drive itself.

level 5 systems have been tested successfully over many years, but while the systems passed tests, the tests also revealed issues----as tests should do.

Some people are saying, "because they aren't here now, they never will be." They Think that makes them sound smart.

Others are saying L5 systems will be a majority in five years .... but ... stupid humans. "Just because it is provably better in every parameter I care about doesn't mean it's better!!" (Ummm ... actually, it does.)

Not to be racial ... but look how long it has taken for the idea that human beings with dark skin and human beings with light skin differ only in skin color. Because by every measure I can see, very, very few people believe that.

What is not to love about a species where people cry and plead for some solution when their children die of rubella and whooping cough and polio ... and a generation later claim that the medicine causes magical brain damage?

And we all know, those are the same people who will pitch the harshest fits when their grandkids get polio.

Shoot ... we live on a planet where the dominant species actively Wants to get involved in wars, and cries about death and murder .... but will not make people take a serious user safety test for either a gun or a car.

L5 cars might be ready for people in five years ... people won't be ready for safe travel for a good while longer I fear .... and still, they will scream and cry every time a stupid driver causes a wreck ... and will also yell at the other driver, when they themselves are the stupid driver who almost causes an accident because they spilled their coffee while driving and texting.

A culling might be in order?

I-Like-To-Bike

Really? Which posters, "people", "others" or "they" here are saying what your straw men are saying?

Rollfast

The autonomous stuff costs enough that they could easily have a fleet three times as large. More cars equal more fares. Add to this mix the fact that Uber is getting booted out of a few PRIME markets now and then, it makes as much sense as caramel covered salami.

Ninety5rpm

Not following.

The autonomous stuff (the radars, lasers, computers, sensors, etc.?) costs enough that they (Uber?) could easily have a fleet three times as large? Three times as large as the fleet they have now? Why would high costs of AV tech allow them to have a fleet three times as large as they have now, especially considering the current fleet is owned entirely by Uber drivers?

More numbers of fares? So they don't need to make as per fare and thus can have lower fares? Is that your point? If not, what is?

The places that booted Uber will have to allow them once the advantages are experienced elsewhere. The economic forces will be irresistible.

Rollfast

The assumption is that these vehicles will become on par pricewise with non-autonomous vehicles, which is not correct because the costs of new safety features always ADD to the price.


Fares have to reflect this. Expenses must be amortised, worked out of them equation quickly to promote profits.


The vehicles are subject to mechanical failures as before and more electronic problems added to the existent ones. Maintenance costs are increased. Once a vehicle is operated, it's all downhill. There is nothing different or magical about them in that respect.


Uber is not interested in replacing all of their fleet with autonomous vehicles by any means. Their business model involves independent contractors performing taxi-type services, with the intent of keeping company expenses lower with little or no owned equipment, only an app and payment, fee distribution guidelines.


Waymo and Lyft are interested in these vehicles, but Lyft is hardly involved to the extent that might seem feasible or likely.


The future of driverless taxis is not here. It's an experiment. In some cases it had a problem or two and was curtailed.


As I've delved into in another thread, this scenario is nowhere near a reality, nor will it be desirable for a large portion of people, hence it is highly premature to theorize about it.


And the reasons they were booted had to do with licensed taxi drivers complaining successfully that Uber violated laws in their area as well as the fact that Uber has behaved rather badly in it's business dealings in general and that alone threatens it as a business in the future. It's not the Wild West, you can't rustle 'horses'.

jefnvk

What about all the driver control systems that can be removed from the vehicle? Steering wheels are not cheap, nor is a steering column. Same goes for traditional linked brake controls. There are savings to be had, not to mention allowing some space to be reclaimed, potentially resulting ins more compact vehicles and less weight which increases fuel economy.

How much will removing the driver save, even if the car does end up costing a bit more? Assume they cost Uber $15/hr, that is $360 per 24h day per vehicle they can save on labor costs. That alone will buy new cars in a hurry.

Yet, you are doing exactly that

Ninety5rpm

The cost of the car is negligible compared to the cost of the human driver, per mile. You can double the cost of the car (which AV tech won't do) and still have far lower costs than human driven hailed cars.

Uber is very interested in owning their own fleets. But I anticipate them to contract with local operators as well. Get a loan, buy a few AVs, contract with Uber, and you're in business. A similar model to an Avis or Enterprise franchise.

The transition will be incremental. Initially customers will have the option to accept a ride from a driverless Uber at a lower rate, but their availability will be limited. As AVailability (see what I did there?) increases, their rates are likely to drop even more, and human drivers will be phased out quickly.

Yes, self-driving cars are still an experiment. But production level is probably only a year or two away. If you doubt this, check out the state of the art 1 1/2 years ago:

Ninety5rpm

It sounds like by 2020 serious deployments will be beginning. Maybe sooner. And it's not just Google/Waymo. There's GM/Cruise, and Aurora already has deals with VW, Hyundai and Byton.

Initially Level 4, but that should be enough for ride hailing within designated metro areas.

Ninety5rpm

Chris Urmson, the former head of Waymo, now CEO of Aurora, is concise about this:


"We believe the actual driving intelligence has to reside on the vehicle. At the end of the day, you can't afford to be in a communication dead spot."

That's why it's dumb.

https://www.autoblog.com/2018/01/17/...omy-interview/

Ninety5rpm

Top 10 AV companies:

1. GM/Cruise (Level 4 Cruise AV without steering wheel to hit the streets in 2019, probably as a Lyft)
2. Waymo
3. Daimler-Bosch (parent of Mercedes Benz)
4. Ford
5. VW (working with Aurora)
6. BMW
7. Aptiv (working with Lyft; purchased nuTonomy)
8. Renault-Nissan-Mitsubishi Alliance
9. Volvo-Autoliv-Ericsson-Zenuity
10. PSA (France)

Dossier: The leaders in self-driving cars | ZDNet

squirtdad

You know, It would really help if you read and comprehended what is posted. No where did it suggest that driving intelligence was not at vehicle level. The company in the link was proposing having remote drivers available for those situations where onboard driving intelligence is not able to resolve a situation, and having the remote driver take control, resolve the situation and then give control back

a simple example of one that I see almost every day is a low speed 4 way stop, where "nice drivers" wave other drivers who have the right of way through. I have not found any reference to AV recognizing a "nice driver wave" as a valid bit of decision information

and for about 6 blocks leading to that 4 way stop the street is so narrow that if cars are parked on either side it is a one way street and drivers have to defer and take turns. I would truly like to see Waymo try their technology in this area.

Ninety5rpm

Really? You think I didn't get that? Please. Give me a break. The point is there are dead spots everywhere and you can't have a system that relies on the ability to communicate with a human operator in real time because having such an ability is not reliable.

The other point is that this is the guy who knows how this stuff works saying this. The technology at Google and now at Aurora can deal with waving drivers; believe me.

Back in 2016 they already could recognize cyclist hand signals:
Google's Self-Driving Cars Can Recognize Cyclist Hand Signals

They are making decisions in fractions of seconds. You can't have these systems throw up their hands, send an SOS message hoping one of the carriers will get the message across, get the video live to the human, wait for the human to comprehend the situation and the problem, and then finally issue commands. It's a really dumb idea. It won't fly. The cars have to be reliable and responsive in real time without this stuff. Certainly for something as mundane as understanding a driver's wave.

jefnvk

I'd love to see us ticket and strip of licenses all those folks who can't be bothered to operate according to the law, creating these scenarios. Arguing that AVs will have an issue because humans can't follow the law is simply a reason we need to remove human drivers.

squirtdad

Do really think that will happen, when we cannot even truly restrict multiple offender drunk drivers, where there are no controls preventing an unlicensed, uninsured driver from operating a vehicle?

beyond that is is not even about who cannot follow a law, in some cases it is really difficult to determine who stopped first at stop sign..... so people hesitate out of safety.

People need to get real about AV in terms of

• current limitations of the technolgy, no matter how cool is its

• limitations of the legal, regulatory, insurance areas
• acceptance by users
• time frame to be broadly used (generational, not years )
• all the areas that it will not be commercially or crtical mass practical
• etc

it will happen, but will not happen as fast or to the extent that marketing types think

jefnvk

No I don't, but its equally silly to say the technology can't work because of faults on the human side. The times I am waved through a stop sign because it was a dead heat v the times I am waved through when it clearly is not my turn is negligible.

As to areas it won't be commercially viable? Is Uber huge right now? Yes. Yet, it is not offered where I am currently at. That doesn't mean it isn't wildly successful, it means that it has found its use case. AVs know what they're targeting, they can easily be successful before we even get to level five.

If I had to side with the timelines of either the marketing folks or the detractors, I'm still siding closer to marketers.

squirtdad

Not saying that technology can't work, just saying there are still some decent size issues and human behavior is not going to change so technology is going have to handle or avoid those situations.

this is a very specific situation with 5 schools, parents taking kids to school, kids walking to school, kids biking to school, adults biking and driving not to school. I see at least 2 - 5 instances a week on the waving through, who is first, whose turn is next on the oneway by default street.

don't disagree, but many suggest AV will replace personal vehicle even in rural, bad weather areas in a few years.

time will tell on timing

350htrr

and... There's the rub. The main reason this AV technology is so far a fail is because of human inter action, where humans are not following the rules... and thus,... the solution will end up to be, to get rid of the human drivers. through a MUCH tougher licencing requirement..

Ninety5rpm

Humans get in the way in two ways.

First, with driver-assist systems (not self-driving), they don't take over when they're supposed to. Level 4 cars without steering wheels will address this.

Second, they have a tendency to rear-end AVs. Oh, well. We've lived with rear-enders this long, a few more decades won't kill us.

I don't see a need to forcefully get rid of human drivers. It will happen organically.

Ninety5rpm

Sounds like Palo Alto where Google cars have been running around for years with nary an incident.

I-Like-To-Bike

All depends on what you consider the appropriate metrics for measuring a "wildly successful" business organization.
Is your idea of wildly successful burning up investors' money at a prodigious rate quarter after quarter, year after year with no light at the end of the tunnel?

Perhaps wildly successfulwill be a payoff through an IPO bonanza for the founders and early investors at the expense of the new shareholders who will be left owning a company with a history of inability/unwillingness to comply with laws and regulations, contempt for its employees and customers, low ethical standards and little hope of ever running a profitable business with or without human drivers.