Car Chat

[giant_mtb]

Infrastructure, weather, and free will. 

Three things that will likely never change enough for truly autonomous vehicles to ever be in everybody's driveway.

[Laconian]

everybody

This is the fallacy. It doesn't have to be for everybody. Only 20% of the US population is rural. If they can deliver a solution that works well for the other 80%, it would be a massive game changer.

How about California or the sunbelt states, where the weather is great and mass transit is nonexistent?

[Soup DeVille]

Also, why is everyone leaning into Autonomous driving? At best, we've got advanced cruise control. None of these systems at the moment are capable of doing anywhere near as much as the stockholders would lead you to believe.


I mean, I have enough trouble with the Lyft driver app; i can't imagine that awful app actually controlling where my car goes.

Whoever gets the software right, and accepted by the buying public first will eventually have a near monopoly on the market, ala Google. They're willing to risk some segments.

[FoMoJo]

Whoever gets the software right, and accepted by the buying public first will eventually have a near monopoly on the market, ala Google. They're willing to risk some segments.

To get it right will be pretty much the same as developing full blown AI. 

[12,000 RPM]

They do seem to be betting on an autonomous future for cars. Think they know something we don't?

After hearing a few Hackett interviews, no, absolutely not.

This is the fallacy. It doesn't have to be for everybody. Only 20% of the US population is rural. If they can deliver a solution that works well for the other 80%, it would be a massive game changer.

How about California or the sunbelt states, where the weather is great and mass transit is nonexistent?

Agreed. Same story with EVs.

The ZDX was and will always be terrible. Huge car, shitty rear passenger compartment made for Japanese-sized bodies.

Most of HMA's fleet can accommodate our corpulent gaijin frames. A new ZDX would be no different.

[Soup DeVille]

To get it right will be pretty much the same as developing full blown AI. 

Oh, no. Traffic is still a limited environment with a finite number of situations.

[MrH]

If a company came out with a fully autonomous living room on wheels (ie no steering wheel, just drives itself) and it worked well, I'd drop serious money on it immediately.

It's a quality of life game changer.

[FoMoJo]

Oh, no. Traffic is still a limited environment with a finite number of situations.

Reading of some of the testing that's going on, and this is Ford and one of their EV partners, the EV was waiting in line to make a left turn at what sounded life a 4 way stop.  The EV kept waiting and waiting as the oncoming traffic stopped waited and then continued.  It seemed that the EV had not pulled up close enough to the line and somehow seemed confused as to when it could go.  They made some adjustments to the software so that the car would stop closer to the line and then take its turn.

It sounded to me like this was a rather small situation in an almost infinite numbers of situations in order for the car to make appropriate decisions.  How's it going to deal with a one lane road when it meets a car coming in the other direction and there's no where to pull over?  What's the other car going to do?  Literally, millions of situations that may only be determined by trial and error.

[Soup DeVille]

Not trial and error, which could lead to some seriously bad decisions, but nearly 100% of scenarios could be simulated, and programming decisions made.

[12,000 RPM]

I think you underestimate the capacity for traffic fuckery. Especially in our increasingly road-ragey distracted driving environment.

[2o6]

I think you underestimate the capacity for traffic fuckery. Especially in our increasingly road-ragey distracted driving environment.

This.




Also, I wouldn't be surprised if the Cruze is just imported from Mexico. GM moves a fair amount of car shapes in Mexico, whereas Ford doesn't.

[FoMoJo]

Not trial and error, which could lead to some seriously bad decisions, but nearly 100% of scenarios could be simulated, and programming decisions made.

No doubt that is true, but I still believe that it would require near AI level software plus infallible sensors and support mechanisms.

[2o6]

Actually, aside from the CT6, Impala, and Volt, did GM actually confirm the deaths of those cars? Or just plant closings?

The Lacrosse could likely be shipped over on a boat, but the Impala is at the end of it's life cycle. The Volt II hasn't been a strong seller, and no one buys Cadillac Sedans. The Cruze can be imported from Mexico, and the Sonic and Spark (if they continue to exist) can be imported from Korea.


This isn't like Ford; who developed a new model cycle of products and didn't consider North America as a whole, and thus decided to concede the market entirely.

[12,000 RPM]

Anything imported is potentially subject to a 25% tariff which basically makes any of these cancelled sedans/hatchbacks too expensive to sell in the US

[2o6]

Anything imported is potentially subject to a 25% tariff which basically makes any of these cancelled sedans/hatchbacks too expensive to sell in the US

Not from Mexico, it's not.



The Lacrosse might get cancelled.

[93JC]

Actually, aside from the CT6, Impala, and Volt, did GM actually confirm the deaths of those cars? Or just plant closings?

GM's announcement didn't say they'd kill any of the product lines off, simply that "future products will be allocated to fewer plants next year", and that Oshawa, Detroit-Hamtramck and Lordstown were being "unallocated".

See https://media.gm.com/media/us/en/gm/news.detail.html/content/Pages/news/us/en/2018/nov/1126-gm.html.

They're also "unallocating" their Baltimore transmission plant (where they build Allison 1000 transmissions for Silverado/Sierras), and the Warren transmission plant in Warren, Michigan (where they build 6T70 & 6T75 transaxles—the GM-Ford 6-speed). Allison already make the 1000 transmission in another plant, and the 6T70/75 are being replaced by the new nine-speed.

[Soup DeVille]

No doubt that is true, but I still believe that it would require near AI level software plus infallible sensors and support mechanisms.

The traditional method used dealing with failsafe systems is to have three identical systems. Any malfunction in any one system is immediately recognized by the other two.

Automotive systems have gone the opposite way so far, using sometimes far too few sensors and little failsafing.

I would expect there to be an eventual compromise and industry standard that resolves most of that concern.

[2o6]

GM's announcement didn't say they'd kill any of the product lines off, simply that "future products will be allocated to fewer plants next year", and that Oshawa, Detroit-Hamtramck and Lordstown were being "unallocated".

See https://media.gm.com/media/us/en/gm/news.detail.html/content/Pages/news/us/en/2018/nov/1126-gm.html.

They're also "unallocating" their Baltimore transmission plant (where they build Allison 1000 transmissions for Silverado/Sierras), and the Warren transmission plant in Warren, Michigan (where they build 6T70 & 6T75 transaxles—the GM-Ford 6-speed). Allison already make the 1000 transmission in another plant, and the 6T70/75 are being replaced by the new nine-speed.

So, as usual, the auto mags have gone off the deep end and aside from a few models that no one liked, (and obviously a lot of jobs lost), really there's no real change.

[Soup DeVille]

I think you underestimate the capacity for traffic fuckery. Especially in our increasingly road-ragey distracted driving environment.

Not at all. My major concern is more of image recognition.

[93JC]

So, as usual, the auto mags have gone off the deep end and aside from a few models that no one liked, (and obviously a lot of jobs lost), really there's no real change.

The auto mags love to jump to conclusions, of course, but it's a pretty good bet that some of the cars are being axed and won't come back. Detroit-Hamtramck and Oshawa are the only plants in the world building Impalas. All Volts are built at Detroit-Hamtramck.

On the other hand the Lacrosse, XTS and CT6 are also built in China; they'll likely stay in production there. Cruze is built in Mexico, China and Argentina; it's also likely it'll stay in production, at the very least in China.

[2o6]

I predict the Volt, XTS and def Impala.


The Cruze will likely shift production, as will the Lacrosse and CT6.


IIRC the Volt's chassis is a compromised design, and is basically a Cruze with a weird shaped battery. Wouldn't be surprised if there's a Volt replacement that's significantly streamlined.

[93JC]

I think it's almost certain that the Impala is dead. I rarely see them out and about, and the handful I have ever seen were all rentals. Same with the Malibu, frankly, but the Malibu seems more popular in the US based on what little I've looked at sales charts. The Impala is just too big, at that size people are flocking to SUVs, and the Malibu isn't much smaller; to me it just doesn't make sense to build the Impala anymore.

I almost never see Lacrosses, XTSes and CT6s either. In fact I don't think I've ever seen a CT6 outside of an auto show. (Speaking of rare GM products I saw my first Buick Regal... "sportback"? Is that what they're calling it? The new liftback. Anyway, saw one today.) It's not surprising that they'd kill these off.

Second-gen. Volts are also very rare here. It's supposedly a very good car, but the hype as compared to the first-gen. is non-existent. I imagine they'll come out with a more comprehensive lineup of electric cars later, to replace the Volt anyway.

[BimmerM3]

This is the fallacy. It doesn't have to be for everybody. Only 20% of the US population is rural. If they can deliver a solution that works well for the other 80%, it would be a massive game changer.

How about California or the sunbelt states, where the weather is great and mass transit is nonexistent?

The other fallacy is the idea that autonomous cars need to be perfect before they reach adoption, when in reality they just have to get to the point where the public trusts them as much or more as other drivers, including the moron watching Netflix on his windshield mounted phone as he cruises down the highway.

[93JC]

Eh, autonomous vehicles are going to have to be pretty much perfect for liability reasons alone. It's not enough to be "as good as the crappy drivers already on the road", the carmakers would be sued into the ground. They already get sued a ton for fairly innocuous product defects, imagine how badly they'd get sued if an autonomous car killed someone.

[Soup DeVille]

Eh, autonomous vehicles are going to have to be pretty much perfect for liability reasons alone. It's not enough to be "as good as the crappy drivers already on the road", the carmakers would be sued into the ground. They already get sued a ton for fairly innocuous product defects, imagine how badly they'd get sued if an autonomous car killed someone.

I feel there's a legal way for the owner to assume liability. Otherwise, yes, you're absolutely right.

[FoMoJo]

The traditional method used dealing with failsafe systems is to have three identical systems. Any malfunction in any one system is immediately recognized by the other two.

Automotive systems have gone the opposite way so far, using sometimes far too few sensors and little failsafing.

I would expect there to be an eventual compromise and industry standard that resolves most of that concern.

Not at all. My major concern is more of image recognition.

Failsafe would be necessary.  Image recognition would be a major concern, but what to do, imo, is the most difficult part.  From the example I mentioned, they had to adjust the software so the car was better positioned for that particular situation.  If they have to programme for every conceivable situation, it would be exhaustive.  I would think that there must be a central AI process that can make common sense decisions based on input.  As well, it would be helpful if cars, within a certain range, could communicate with each other.

[giant_mtb]

This is the fallacy. It doesn't have to be for everybody. Only 20% of the US population is rural. If they can deliver a solution that works well for the other 80%, it would be a massive game changer.

How about California or the sunbelt states, where the weather is great and mass transit is nonexistent?

What happens when the gub'ment decides to make human-driven vehicles illegal or prohibitively expensive because they're "dangerous." 

[93JC]

I feel there's a legal way for the owner to assume liability. Otherwise, yes, you're absolutely right.

I'm sure the legal departments at the many companies working on autonomous vehicles have been busy trying to devise an End-User Licence Agreement that would shift liability to the buyer, but I doubt that would hold up in a court.

[93JC]

Very prescient article from an Ontario newspaper: https://www.thepeterboroughexaminer.com/news-story/8210827-gm-oshawa-will-shut-down-by-2019-consultant/

Published five years ago, mind you. This guy definitely saw the writing on the wall for the Oshawa plant.

[FoMoJo]

Very prescient article from an Ontario newspaper: https://www.thepeterboroughexaminer.com/news-story/8210827-gm-oshawa-will-shut-down-by-2019-consultant/

Published five years ago, mind you. This guy definitely saw the writing on the wall for the Oshawa plant.

It would make sense for someone like Hyundai to take over the Oshawa facilities.  Currently, they are selling well here and have no manufacturing facilities.

GM has been a bad partner in Ontario for years, threatening to pull out, accepting bailouts and, finally, just walking away.