Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
PenisMonkey
Apr 30, 2004

Be gentally.
I’m surprised that person could even get FSD to engage in that weather.

https://i.imgur.com/MWOrDTT.mp4

Adbot
ADBOT LOVES YOU

Nfcknblvbl
Jul 15, 2002

Tesla should watermark their dashcam footage to indicate whether it's in AP/EAP/FSD.

WhiteHowler
Apr 3, 2001

I'M HUGE!

Rusty posted:

Multiple people have said it turns in front of cars at left turn lanes, why does this stuff have no regulation at all? Do other countries also have this live beta test on their roads?

Unfortunately, the only thing likely to put a spotlight on this will be a high-profile fatality caused by self-driving, some celebrity or congressperson's kid or something. It will happen sooner or later.

In the meantime, all us normal folks can do is drive extra-defensively.

Tiny Timbs
Sep 6, 2008

Tom Guycot posted:

Lol the rest of that thread is a trip:

https://twitter.com/WholeMarsBlog/status/1595130005893173248




"It will try to kill other people, but have fun! its safer than driving yoruself!"

If the automation is jerking your car and putting it into an unsafe situation before anybody could possible react I don't see how it's helpful to tell people they just need to pay attention. Sure, that's great for dealing with the aftermath but not every bad move by FSD is going to come with a warning it's about to happen.

Tiny Timbs fucked around with this message at 21:40 on Nov 28, 2022

pun pundit
Nov 11, 2008

I feel the same way about the company bearing the same name.

Tiny Timbs posted:

If the automation is jerking your car and putting it into an unsafe situation before anybody could possible react I don't see how it's helpful to tell people they just need to pay attention. Sure, that's great for dealing with the aftermath but not every bad move by FSD is going to come with a warning it's about to happen.

Having used adaptive cruise in a TM3 with HW2, I've learned to watch out for stuff that will make the adaptive cruise do phantom braking. It's still less tiring than constantly adjusting my speed to that of traffic. I wouldn't use self driving though.

I guess after the car has tried to kill you ten or twenty times you will learn its quirks and be about as safe during FSD as if you had a bad tempered dog loose in the passenger seat. You'll be able to predict when it's going to jump into your lap and bark at oncoming traffic, and take steps to mitigate the consequences.

bad_fmr
Nov 28, 2007

PenisMonkey posted:

I’m surprised that person could even get FSD to engage in that weather.

https://i.imgur.com/MWOrDTT.mp4

Looks like normal rain?

Tiny Timbs
Sep 6, 2008

pun pundit posted:

Having used adaptive cruise in a TM3 with HW2, I've learned to watch out for stuff that will make the adaptive cruise do phantom braking. It's still less tiring than constantly adjusting my speed to that of traffic. I wouldn't use self driving though.

I guess after the car has tried to kill you ten or twenty times you will learn its quirks and be about as safe during FSD as if you had a bad tempered dog loose in the passenger seat. You'll be able to predict when it's going to jump into your lap and bark at oncoming traffic, and take steps to mitigate the consequences.

Yeah and similarly for my Polestar 2 the manufacturer warns you that the lane centering likes to jerk towards lane exits along with a couple other conditions you have to keep in mind to use it safely. That's not ideal, but knowing the conditions under which the automation fails does a lot to help people avoid issues. With FSD it sounds like people think it's reasonable to say "well just stay frosty" because the automation can just do whatever whenever, and then they blame people for whatever consequences come out of getting a 100% steering input applied within 0.1s as if they're supposed to have any control over that.

Tiny Timbs fucked around with this message at 22:16 on Nov 28, 2022

YOLOsubmarine
Oct 19, 2004

When asked which Pokemon he evolved into, Kamara pauses.

"Motherfucking, what's that big dragon shit? That orange motherfucker. Charizard."

WhiteHowler posted:

I love the idea of autonomous driving, and once it's tested, refined, and widespread, it will absolutely save a ton of lives.

In the 2030's, probably.

A bunch of companies working on autonomous driving have given up or significantly scaled back their ambitions. Turns out simulating human intelligence is really hard and simply throwing ML at it isn’t a suitable solution.

I doubt any of us will see L5 autonomy in our lifetimes.

Duck and Cover
Apr 6, 2007

I'd like automation to exist but the barrier is going to be "way better than self driving also it has to work with people self driving and no we won't be increasing our infrastructure costs by changing anything ". Which while understandable does make things harder.

Indiana_Krom
Jun 18, 2007
Net Slacker

This person hydroplaned, the "sudden jerk" was probably hitting a puddle. Also they were on a freeway which means no FSD because it automatically switches to navigate on autopilot in the current software stack. Yeah, they were probably driving too fast for conditions, and/or their tires were too worn to be used in that kind of weather. There are countless examples of Teslas and other vehicles spinning out the same way in the same conditions. It would have done it at that speed even if they weren't using any sort of driver assist technology, ultimately the driver is the one who set the speed. Autopilot/FSD currently have no allowances for bad weather or low traction, it just goes however fast you have it set to go, no matter how stupid that speed might be (as long as its under 80 MPH right?). The software isn't some magic that will keep your vehicle on the road even when you are sending it at high speed in heavy rain which means your vehicle is literally not touching the road anymore.

Edit: Not saying autopilot couldn't have handled this better (it could have slowed down automatically before it hydroplaned), but this person clearly needs actual level 5 FSD that would have slowed down for conditions because they obviously don't know how to drive.

Indiana_Krom fucked around with this message at 23:26 on Nov 28, 2022

bird with big dick
Oct 21, 2015


idgi is he saying it was lucky that he was alert because his alertness allowed him to expertly spin his car into the ditch?

priznat
Jul 7, 2009

Let's get drunk and kiss each other all night.

bird with big dick posted:

idgi is he saying it was lucky that he was alert because his alertness allowed him to expertly spin his car into the ditch?

360 n0scope :smuggo:

Saukkis
May 16, 2003

Unless I'm on the inside curve pointing straight at oncoming traffic the high beams stay on and I laugh at your puny protest flashes.
I am Most Important Man. Most Important Man in the World.

dissss posted:

I don't get what the difference between Skoda and VW models is supposed to be these days - not much difference on the ICE models and even less on their EV ranges

I haven't done serious comparison, I've never considered either. As someone who at least reads car magazines my impressions are that they are basically identical in drivetrains and other features that wouldn't matter to customers, but there are enough visible differences to draw customers to either. If someone is enamoured by shiny touch buttonx you sell them a VW, but if they know why touch buttons suck you sell them a Škoda. Like Škoda's Simply Clever features. All cars should have wiper fluid tank cap that forms a funnel, or a USB-C port for your dash cam next to the rear view mirror. Yet Volkswagen doesn't seem to have those and Škoda can use them in their marketing. Both also have some models that are about different enough you could say the other brands don't have comparable model.

Another impression I have is that Škoda is as good as Volkswagen but slightly cheaper, so you could ask why would anyone buy VW when Škoda exists.

It is largely badge engineering so you could if there is a point for all the different brands, but I can think of at least three justifications.

Volkswagen still seems to have slightly more brand cachet, so you can make Germans pay a little bit extra.

VAG also has such a large market share you would probably run into a limit how many you could actually sell as Volkswagens. There would be so many VWs around many would rather buy a Renault.

Also if you tried to make the Czech Škoda workers to build Volkswagens alone they might not be willing to do quite as good of a job.

bad_fmr
Nov 28, 2007

Indiana_Krom posted:

This person hydroplaned, the "sudden jerk" was probably hitting a puddle. Also they were on a freeway which means no FSD because it automatically switches to navigate on autopilot in the current software stack. Yeah, they were probably driving too fast for conditions, and/or their tires were too worn to be used in that kind of weather. There are countless examples of Teslas and other vehicles spinning out the same way in the same conditions. It would have done it at that speed even if they weren't using any sort of driver assist technology, ultimately the driver is the one who set the speed. Autopilot/FSD currently have no allowances for bad weather or low traction, it just goes however fast you have it set to go, no matter how stupid that speed might be (as long as its under 80 MPH right?). The software isn't some magic that will keep your vehicle on the road even when you are sending it at high speed in heavy rain which means your vehicle is literally not touching the road anymore.

Edit: Not saying autopilot couldn't have handled this better (it could have slowed down automatically before it hydroplaned), but this person clearly needs actual level 5 FSD that would have slowed down for conditions because they obviously don't know how to drive.

I have hydroplaned countless times and otherwise lost traction on snow and ice and never has it caused an instant spin on a straight road. In a corner sure the rear may step out. But in a straight line no. That spin is for sure caused by the autopilot. It either applied power/brakes incorrectly or made sudden steerin motion that caused the spin.

dissss
Nov 10, 2007

I'm a terrible forums poster with terrible opinions.

Here's a cat fucking a squid.

Saukkis posted:

It is largely badge engineering so you could if there is a point for all the different brands, but I can think of at least three justifications.

Volkswagen still seems to have slightly more brand cachet, so you can make Germans pay a little bit extra.

VAG also has such a large market share you would probably run into a limit how many you could actually sell as Volkswagens. There would be so many VWs around many would rather buy a Renault.

Also if you tried to make the Czech Škoda workers to build Volkswagens alone they might not be willing to do quite as good of a job.

Maybe all of that is true in Europe, but I'm in a tiny market (New Zealand) and we have VW, Skoda and SEAT.

There is almost 100% overlap between the three brands, only real difference is SEAT markets their more performancey models as Cupra

Russian Bear
Dec 26, 2007


:lol: at trusting anything automated (even old rear end regular cruise control) in adverse road conditions. I hate this version of reality.

bad_fmr
Nov 28, 2007

VW: more posh with luxuries such as capasitive buttons
Seat: more edgy for the younger buyer (lol)
Skoda: more practical and usually a bit roomier interiors, for the young family with small children and a dog

bird with big dick
Oct 21, 2015

bad_fmr posted:

I have hydroplaned countless times and otherwise lost traction on snow and ice and never has it caused an instant spin on a straight road. In a corner sure the rear may step out. But in a straight line no. That spin is for sure caused by the autopilot. It either applied power/brakes incorrectly or made sudden steerin motion that caused the spin.

Or the driver that we already have proof of being a lovely driver applied power/brakes incorrectly or made a sudden steering motion.

CAT INTERCEPTOR
Nov 9, 2004

Basically a male Margaret Thatcher

bad_fmr posted:

I have hydroplaned countless times and otherwise lost traction on snow and ice and never has it caused an instant spin on a straight road. In a corner sure the rear may step out. But in a straight line no. That spin is for sure caused by the autopilot. It either applied power/brakes incorrectly or made sudden steerin motion that caused the spin.

Personal experience does not negate that hydroplaning on a highway in a straight line does in fact occur. That spin is very much like a hydroplane. Now wether automation helped it along is a pretty good question and one I'm inclined to believe as it's not unknown for cruise controls to misbehave in the wet. A optical style automation like Tesla's FSD moreover could get tricked easier to make the situation worse like Subaru's Eyesight can. The difference might be Eyesight will shut down if it has issues with its sensors and data, is that the case with FSD?

Eyesight does also have a phantom braking issue. Another difference here is Eyesight does not have steering authority over the driver like FSD presumes so my thought would be yes its a hydroplane induced spin, a reasonable argument can be made FSD helped to make the spin unrecoverable.

But the Internet being as it is and a lot of people willing to pronounce every Telsla crash FSD`s fault, I'll have a healthy amount of scepticism - however I'll 100% agree that FSD should not be allowed esp as an admitted beta and where the hell are the regulators to stop it? Beta software has no place on the roads and should not be in the hands of regular drivers.

bad_fmr
Nov 28, 2007

CAT INTERCEPTOR posted:

Personal experience does not negate that hydroplaning on a highway in a straight line does in fact occur. That spin is very much like a hydroplane. Now wether automation helped it along is a pretty good question and one I'm inclined to believe as it's not unknown for cruise controls to misbehave in the wet. A optical style automation like Tesla's FSD moreover could get tricked easier to make the situation worse like Subaru's Eyesight can. The difference might be Eyesight will shut down if it has issues with its sensors and data, is that the case with FSD?

Eyesight does also have a phantom braking issue. Another difference here is Eyesight does not have steering authority over the driver like FSD presumes so my thought would be yes its a hydroplane induced spin, a reasonable argument can be made FSD helped to make the spin unrecoverable.

But the Internet being as it is and a lot of people willing to pronounce every Telsla crash FSD`s fault, I'll have a healthy amount of scepticism - however I'll 100% agree that FSD should not be allowed esp as an admitted beta and where the hell are the regulators to stop it? Beta software has no place on the roads and should not be in the hands of regular drivers.

Hydroplaning does happen, thats obvious. But hydroplaning on a straight line by itself does not cause that kind of violent spin. You need steering, brake or acceleration input for that to happen. So yes it is debatable that the car was losing traction and then the autopilot poo poo the bed causing the crash by wrong inputs, or the autopilot caused the loss of traction in the first place by ghost braking/steering or whatever. Neither is very good option.

gwrtheyrn
Oct 21, 2010

AYYYE DEEEEE DUBBALYOO DA-NYAAAAAH!

bad_fmr posted:

VW: more posh with luxuries such as capasitive buttons

the "luxury" of capacitive buttons :barf:

CAT INTERCEPTOR
Nov 9, 2004

Basically a male Margaret Thatcher

bad_fmr posted:

Hydroplaning does happen, thats obvious. But hydroplaning on a straight line by itself does not cause that kind of violent spin. You need steering, brake or acceleration input for that to happen

That is not correct - there are other factors than may cause an out of control hydroplane like that where neither of those three are required. Tyre condition is the big one, road surface / white lines may also cause unequal traction, as well as the possibility that the hydroplane caused the front wheels to deviate from straight ahead. A lack of input can also cause the deviation from straight as well, rapidly becoming out of control for an inattentive driver. Also water affecting one side of the vehicle can cause a violent rotation with the sudden increase of drag applied to one side only.

It's often the one side hydroplane where cruise controls and ESC can really gently caress things up as the inputs suddenly do not see data that is appropriate to control the vehicle appropriately and react badly. This is not especially a FSD issue and nor is it just a optical ACC exclusive problem, it is often recommended by manufactures to not use ACC / normal cruise control in adverse conditions. So thats why while I dont believe the initial hydroplane loss of control was FSD initiated, I can believe it made the situation worse - however thats not just a FSD issue, but a wider problem with ACC and ESD. Still, given FSD is a beta product as well, that really doesnt sit well with me that it's in use by the average driver in adverse conditions. I've been a semi professional driver trainer, the average driver is loving *terrible* at responding when electronics go nuts and honestly beta automation software in the average driver's hands is just a whole bunch of gently caress riiiiight off.

bad_fmr
Nov 28, 2007

CAT INTERCEPTOR posted:

That is not correct - there are other factors than may cause an out of control hydroplane like that where neither of those three are required. Tyre condition is the big one, road surface / white lines may also cause unequal traction, as well as the possibility that the hydroplane caused the front wheels to deviate from straight ahead. A lack of input can also cause the deviation from straight as well, rapidly becoming out of control for an inattentive driver. Also water affecting one side of the vehicle can cause a violent rotation with the sudden increase of drag applied to one side only.
Well yes, this is all true but it doesnt really contradict what I said. For the unequal traction to cause a spin it still requires some change of direction or speed to cause spinning. If you are coasting on neutral straight ahead it does not really matter what the traction is on either side, right? Only when you brake or hit the gas the difference in traction causes the spin. Normally this is kind of situation is checked and corrected by the ESC systems anyway, of course only up to a point.

Obviously if you lose all grip then the car may start to rotate, but not in the way as in the video. Think more like a curling stone slowly turning.

CAT INTERCEPTOR posted:

It's often the one side hydroplane where cruise controls and ESC can really gently caress things up as the inputs suddenly do not see data that is appropriate to control the vehicle appropriately and react badly. This is not especially a FSD issue and nor is it just a optical ACC exclusive problem, it is often recommended by manufactures to not use ACC / normal cruise control in adverse conditions. So thats why while I dont believe the initial hydroplane loss of control was FSD initiated, I can believe it made the situation worse - however thats not just a FSD issue, but a wider problem with ACC and ESD. Still, given FSD is a beta product as well, that really doesnt sit well with me that it's in use by the average driver in adverse conditions. I've been a semi professional driver trainer, the average driver is loving *terrible* at responding when electronics go nuts and honestly beta automation software in the average driver's hands is just a whole bunch of gently caress riiiiight off.
If cruise or ACC would be causing these kinds of accidents in bad conditions I dont believe they would be allowed on the road at all. Somehow they seem to be working just fine even in winter and icy conditions. The only issues people are complaining is that snow fucks up the ACC radars and then it only works as a standard CC until cleaned.

PenisMonkey
Apr 30, 2004

Be gentally.
As was pointed out, FSD does not engage on a freeway. This was not FSD.

QuarkJets
Sep 8, 2008

bad_fmr posted:

Hydroplaning does happen, thats obvious. But hydroplaning on a straight line by itself does not cause that kind of violent spin. You need steering, brake or acceleration input for that to happen. So yes it is debatable that the car was losing traction and then the autopilot poo poo the bed causing the crash by wrong inputs, or the autopilot caused the loss of traction in the first place by ghost braking/steering or whatever. Neither is very good option.

That's my thought as well, I've hydroplaned while driving in a straight line before but did not suddenly go into a fast spin. I guess you could hit something in the road to gain that angular momentum but that doesn't look to be the case here

e: I could totally believe that an ECS fault caused the spin, which isn't any better than FSD causing it

QuarkJets fucked around with this message at 11:16 on Nov 29, 2022

knox_harrington
Feb 18, 2011

Running no point.

bad_fmr posted:

Well yes, this is all true but it doesnt really contradict what I said. For the unequal traction to cause a spin it still requires some change of direction or speed to cause spinning. If you are coasting on neutral straight ahead it does not really matter what the traction is on either side, right? Only when you brake or hit the gas the difference in traction causes the spin. Normally this is kind of situation is checked and corrected by the ESC systems anyway, of course only up to a point.

Obviously if you lose all grip then the car may start to rotate, but not in the way as in the video. Think more like a curling stone slowly turning.

If cruise or ACC would be causing these kinds of accidents in bad conditions I dont believe they would be allowed on the road at all. Somehow they seem to be working just fine even in winter and icy conditions. The only issues people are complaining is that snow fucks up the ACC radars and then it only works as a standard CC until cleaned.

Excuse mee I think you'll find Cat Interceptor is a High Performance Driver in a dry country, which trumps your personal experience of driving in the wet

Mr Darcy
Feb 8, 2006
Got my Skoda this morning.

Not had chance to do much more than park it in the driveway yet. Work apparently prefer me to actually be online and working during the day.

Let's see if I've screwed up attaching images again.

Edit: possibly yes. Image looks as fuzzy as hell. Direct link here:
https://imageshack.com/i/pmjVz48Zj

Mr Darcy fucked around with this message at 12:19 on Nov 29, 2022

bad_fmr
Nov 28, 2007

Mr Darcy posted:

Got my Skoda this morning.

Not had chance to do much more than park it in the driveway yet. Work apparently prefer me to actually be online and working during the day.

Let's see if I've screwed up attaching images again.

Edit: possibly yes. Image looks as fuzzy as hell. Direct link here:
https://imageshack.com/i/pmjVz48Zj





Nice! Congratulations.
Enyaq would be my EV choice too if I had to buy one today.

pun pundit
Nov 11, 2008

I feel the same way about the company bearing the same name.

I've hydroplaned in a straight line and had spin induced. I was driving in gear, though, rather than coasting in neutral as it's suggested everyone does when driving in the wet in this thread. I don't understand how you would get anywhere like that, tbh.

bird with big dick
Oct 21, 2015

bad_fmr posted:

Well yes, this is all true but it doesnt really contradict what I said. For the unequal traction to cause a spin it still requires some change of direction or speed to cause spinning. If you are coasting on neutral straight ahead it does not really matter what the traction is on either side, right? Only when you brake or hit the gas the difference in traction causes the spin. Normally this is kind of situation is checked and corrected by the ESC systems anyway, of course only up to a point.

Obviously if you lose all grip then the car may start to rotate, but not in the way as in the video. Think more like a curling stone slowly turning.

If cruise or ACC would be causing these kinds of accidents in bad conditions I dont believe they would be allowed on the road at all. Somehow they seem to be working just fine even in winter and icy conditions. The only issues people are complaining is that snow fucks up the ACC radars and then it only works as a standard CC until cleaned.

Lmao cruise doesn’t cause accidents despite it being common knowledge that you don’t use cruise in inclement weather because it can cause an accident.

quote:

only when you brake

What happens in most EVs when you take your foot off the gas?

Please tell me how you’ve completely eliminated “driver is a dipshit that jerked the wheel when he got scared when his car started hydroplaning” when it has already been literally proven that the driver is a dipshit due to him using AP in those conditions when he shouldn’t have been.

You don’t what you’re talking about and just have a hate boner for tesla.

bird with big dick fucked around with this message at 16:31 on Nov 29, 2022

Bandire
Jul 12, 2002

a rabid potato

Yeah just from what we can see, the car jerks a tiny bit left before it goes hard right. My money is on him hitting a little standing water pooling on the left side of the roadway pulling him toward the HOV poles, and then he freaked out and jerking the wheel to the right. He could have very easily assumed that pull to the left was the car and not a puddle.

bird with big dick
Oct 21, 2015

pun pundit posted:

I've hydroplaned in a straight line and had spin induced. I was driving in gear, though, rather than coasting in neutral as it's suggested everyone does when driving in the wet in this thread. I don't understand how you would get anywhere like that, tbh.

Now everyone ask themselves if EVs with regen, especially 1PD style regen, are more like driving in gear or more like driving in neutral.

ESC cannot save you when you have nearly zero traction from hydroplaning.

I spun out exactly like this just due to engine braking when I came out of the Eisenhower tunnel and hit unexpected snow. No AP required.

bad_fmr
Nov 28, 2007

Is it really controversial in this thread to say that for cars to spin when losing traction there generally needs to be a source of torque and the most common ways to provide that torque is by accelerating, braking or turning the wheels? And the best way to recover from the loss of traction is to ease off the gas and to not touch the brakes, to prevent the spin? This is like the very first thing you learn on any ice driving or slippery driving school.

:psyduck:

I do not know who gently caress up in the video the driver or the car or both. It was labelled as autopilot error so that was the assumption. My point was that the car in question does not just spin because of some hydro. After it loses traction (if it is what happens) someone or something fucks it up causing it to spin.

bird with big dick posted:

You don’t what you’re talking about and just have a hate boner for tesla.

:jerkbag:

bird with big dick
Oct 21, 2015

bad_fmr posted:

That spin is for sure caused by the autopilot.


quote:

I do not know who gently caress up in the video the driver or the car or both.

:jerkbag:

wolrah
May 8, 2006
what?

Wizard of the Deep posted:

This is AMERICA, where we have freedom to be free in testing new technology! Regardless of how dangerous it is.
The biggest thing as I see it is they've managed to get everyone referring to this as beta testing. No, average idiots aren't loving testing anything. They're using incomplete and known buggy software in a safety critical application, that's it. Testing involves procedures, repetition, meaningful communication between the testers and the developers, etc. Tesla's public bullshit isn't any of that, and there's little evidence that they're even doing much of that internally.

Given the nerd-heavy demographics of this forum I'm sure I'm far from the only one who has been involved in multiple actual beta tests, where we had specific targets of things to test, tools to generate debug logs when things went wrong, and a ticket system where those reports were tracked, classified, and responded to by developers who could then ask us questions and have us retest after something was changed until it worked as expected.

What Tesla is doing is the approach made popular by Google of shipping as soon as it works well enough to demo and just using the beta label to excuse the bugs. That's fine for an email client, mapping software, etc. where inconvenience or minor personal data loss are the worst case scenarios, but absolutely insane for a safety critical application where not only the user but unwilling people around them are at risk.

YOLOsubmarine posted:

A bunch of companies working on autonomous driving have given up or significantly scaled back their ambitions. Turns out simulating human intelligence is really hard and simply throwing ML at it isn’t a suitable solution.

I doubt any of us will see L5 autonomy in our lifetimes.
Agreed, I have yet to see anything that makes me believe full L5 autonomy, as in drive anywhere a reasonable human would drive a comparable vehicle, is even possible without actual sci-fi grade general AI. Even if that can be technically achieved, which is very much an open question, it'd be an entire moral can of worms.

Very good L4 is definitely possible, and lots of progress has been made in that realm over the last decade, but I'm doubtful it will ever become mainstream in personally owned vehicles. Local taxi, transit, and delivery applications, sure, but I wouldn't hold my breath for anything better than L3 in the kinds of cars normal people park in their garages.

bad_fmr posted:

Is it really controversial in this thread to say that for cars to spin when losing traction there generally needs to be a source of torque and the most common ways to provide that torque is by accelerating, braking or turning the wheels? And the best way to recover from the loss of traction is to ease off the gas and to not touch the brakes, to prevent the spin? This is like the very first thing you learn on any ice driving or slippery driving school.
Have you ever intentionally driven in to a puddle at speed and hydroplaned just one side of the car? The combination of power loss on that side plus the resistance of the water can create some pretty strong rotational forces. Split mu situations are sketchy even when you're expecting it.

Three Olives
Apr 10, 2005

Don't forget Hitler's contributions to medicine.

wolrah posted:

Agreed, I have yet to see anything that makes me believe full L5 autonomy, as in drive anywhere a reasonable human would drive a comparable vehicle, is even possible without actual sci-fi grade general AI. Even if that can be technically achieved, which is very much an open question, it'd be an entire moral can of worms.

This is a thing that I can't tell if the L5 people are lying about or just, naïve about, what exactly problem they are working on and it's scale.

Everyone talks about L5 like it is video game, driving however requires understanding how the world actually works on a higher level, basically, understanding theory of mind for things they interact with, being able to deduce the intent, capabilities and limitations of other objects and how they interact with each other.

Like, it's not enough to know it is safe to cross under a bridge with train tracks on it, you also need to understand how a train on the bridge is different than a train on a bridge full of burning chemical tanks. It's not even enough to recognize a train on fire, you need to understand fire, what it does, what kinds of dangers it poses.

You can't teach every edge case, say you are like me and see a wheel rolling down the interstate not attached to a car, it's not enough to understand that you need to avoid the wheel, you need to understand that that wheel was attached to a car and if it is rolling down the interstate at 60 miles an hour by itself something really bad has happened and you need to anticipate that a very bad car accident has just happened that you are approaching at 65 MPH.

Russian Bear
Dec 26, 2007


Cars, autonomous or otherwise need be a small piece of the transportation future. Like low single digit %. Spending a ton of money to get to L5 is a fool’s errand.

Edward IV
Jan 15, 2006

On a different note, I rented a ID.4 (AWD Pro S) over the weekend and I like it to the point that I wish I could have kept it. My only complaint is that getting out is a little awkward particularly trying to contort out of the seat to get my legs out of the car. I don't know if it's just the seat bolsters are too conforming and not broken in enough to twist my rear end out or if I just need a new technique and use the grab handle to lift myself out first. It's a bigger car than I'm used to although it handles just as well as my Impreza and Forester probably thanks to the low center of gravity. I do kind of wish the cheaper AWD Pro was still available for reservation (yes I know that RWD with proper winter tires are better in snow than AWD with all-seasons) because it does scoot even in Eco mode and it will hurt the resale value if I try to trade it in or sell it here in NJ. The AWD Pro S is just too much money and cuts out half of the NJ EV incentives.

I didn't dig into the infotainment system that deeply as I mainly relied on Android Auto. It and the touch controls were tolerable which is probably the best thing anyone could say about it. I definitely did not feel comfortable playing with it while driving which I know I shouldn't be doing in the first place but it really exacerbates the problem of taking your eyes and attention off of the road.

I also charged it at my apartment and it worked technically. However, the included 120V charger was not sufficiently long enough to easily reach into my apartment so I had to use an extension cable which was definitely sketchy. Especially because the charger plug and box ended up being outdoors which definitely would be a very bad idea if the weather were to get wet. There is precisely only one parking spot where I can place the charger box indoors albeit sitting right on the window sill. I didn't think to check to measure the length of the included charging cable to see if I could get anything longer. The room closest to the parking lot (which happens to be my bedroom) does have a 6-20 socket for an AC unit but the AC unit installed runs on 120V so it's unoccupied right now so I could use it. However, that 240V circuit is only on a 15A breaker so I'm stuck with getting a low amperage charger (<12A I guess) or use the low current charging on the ID.4 which I think limits it to 10A. The improvement in charging rate would be only marginal but at least the circuit would be isolated from everything else in the apartment. I actually ended up tripping a 20A breaker because the charger and a portable induction cooktop ended up being on the same circuit. In any case, I'm going to need as long of a charging cable that I can get so the plug and box can stay indoors and have some flexibility of where to hook it up to not overload a circuit.

I'll also need to check with the super and property manager if this would fly long term because I could see their point that running a power cable out my window into the parking lot being a nuisance. Also need to see if they would be willing to install some kind of charging infrastructure although I could see issues as to how to check who to bill for power usage. I do have some other free options including free L2 charging at a local Stop & Shop and 30 minutes of free DC charging at one of the strip malls near me. Fortunately, my commuting and driving needs won't require me to be charging super often. Going from 45% to 80% did need about 24 hours on 120V and I estimate using the 6-20 socket at 10A would only drop that to about 19 hours. I have until February at the earliest to get this all sorted out.

Edward IV fucked around with this message at 20:45 on Nov 29, 2022

CAT INTERCEPTOR
Nov 9, 2004

Basically a male Margaret Thatcher

bad_fmr posted:

Well yes, this is all true but it doesnt really contradict what I said. For the unequal traction to cause a spin it still requires some change of direction or speed to cause spinning. If you are coasting on neutral straight ahead it does not really matter what the traction is on either side, right? Only when you brake or hit the gas the difference in traction causes the spin. Normally this is kind of situation is checked and corrected by the ESC systems anyway, of course only up to a point.

Obviously if you lose all grip then the car may start to rotate, but not in the way as in the video. Think more like a curling stone slowly turning.


You havent accounted for the fact that water has it's own weight and applies forces back to the car that can be considerable - and can exert forces beyond what just the brakes or throttle / steering inputs can do. Hydroplaning isnt just about losing grip in isolation, hitting a standing body of water can provoke a violent reaction esp if the water has any depth or the water is not square to the car so that it is possible one or two wheels are not actually in contact with the road but also have a considerable amount of drag - or worse, one or two wheels are having considerable drag while the other wheels lose grip. All of that combined does unsettle cars and can do so with some violence and speed. Also factor in other issues like road paint has it's own change in grip levels, road surface may change and it's even more possible that a car may quite abruptly lose control in adverse conditions.


knox_harrington posted:

Excuse mee I think you'll find Cat Interceptor is a High Performance Driver in a dry country, which trumps your personal experience of driving in the wet

The country thats been flooded for the last three years?

Adbot
ADBOT LOVES YOU

mAlfunkti0n
May 19, 2004
Fallen Rib

Russian Bear posted:

Cars, autonomous or otherwise need be a small piece of the transportation future. Like low single digit %. Spending a ton of money to get to L5 is a fool’s errand.

I couldn’t agree more. I wish trains were actually a thing here in the US. Eddie’s it being a “they’re always late and slow” thing.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply