Driverless Audi RS 7 at racing speeds at Hockenheim

[RC45]

[citation needed]



'Odd scenario' can accurately describe the failure of any of the other thousands of parts of a car that could malfunction at any point.



What are you, allergic to reading? Automation isn't about eliminating human error, it's about minimizing error, period.

No, you still haven't addressed this. Either start backing up your ridiculous claims, or stop making them.

Video proof already provided. Ice Mode Initiated.

If automated vehicles are intended to minimize human error, why pass control to the error prone human when the automated system suffers a sensory failure? What if the human is not ready to take over again? If the human is required to sit at the ready on guard to take over when the automated system suffers a sensory failure, why have the automated system in the first place?

Thank you again for underlining the fact that your proposed automated car is a solution looking for a problem.

 

[Tornado]

Forget unfortunate predicament - what about under normal circumstances if the automated car suffers a sensory failure?

The same thing any non automated car built in the past 18 years responds to a sensor failure: Warns the driver and requires him to respond as necessary.

After all, even at norm al speeds the car is deaf, dumb and blind if it suffers a sensory input failure and all the logic in the world cannot help an input less computer.

Interesting how a single sensory failure would completely cripple the entire computer system of the theoretical car.

 

[Bopop4]

Interesting how a single sensory failure would completely cripple the entire computer system of the theoretical car.

The rear windshield wiper park function on my car stopped working, the wheels then fell off and the gas tank exploded.

I'd call that proof enough.

 

[RC45]

The same thing any non automated car built in the past 18 years responds to a sensor failure: Warns the driver and requires him to respond as necessary.

So the entire automated freeway commuter infrastructure now relies on those pesky humans again.

Interesting how a single sensory failure would completely cripple the entire computer system of the theoretical car.

What happens to a current model automobile with Tyre Pressure Monitors that suffers a TPM failure? That's a single sensor failure. (Pro-tip - this system is government mandated and is required to behave in a certain way of it fails)

What happens to a current model automobile that suffers a throttle position sensor or throttle blade position sensor failure? That's a single sensor failure. (pro-tip - this system must respond in a certain way to help save you and the engine if it fails)

What happens to a current model automobile that suffers a steering wheel position sensor failure? That's a single sensor failure. (pro-tip - this sensory input is used by a number of different systems to compare actual to intended direction of travel and input)

What happens to a current model automobile that suffers a wheel speed sensor failure? That's a single sensor failure.

3 of the 4 scenarios will induce a reduced power/shutdown mode - the 1 of the 4 will shutdown one of the aid systems and if another system should suffer a sensory input failure will shut the car down.

With the automated car we are no longer just talking about passive safety response systems or engine management systems - we are talking about proximity sensors, distance sensors, closing rate sensors, speed sensors - systems used to place the car in a known position on the ground to help navigate obstacles etc.

So, what happens when 1 of those sensors fail? You don't think it will 'shut down entire subsystems'?

If not, then I sure as hell do not want you on any of the engineering and design teams.

Man - these sensory input based systems sure get complicated quickly, don't they?

But then what would I know - I am just an Internet wannabe right?

 

[SlipZtrEm]

Whatever tone you want to assign to my posts, feel free. I do find it strange you think the AUP is below you, though: so once more, with feeling.

Video proof already provided. Ice Mode Initiated.

Video of a car on a track. Not the road-going scenario (nor speed) your hypothetical situation originally alluded to. Not that that matters; just like you've done numerous times in this thread, if evidence is provided that doesn't align with your example, you'll just move the goalposts to another equally-unlikely hypothetical.

If automated vehicles are intended to minimize human error, why pass control to the error prone human when the automated system suffers a sensory failure?

You're... you're serious?

Oh.

Automated systems are intended to minimize human error, yes. Glad you're finally on board. Why pass control back to the human if there's a sensory failure? Because the other option is to have no party controlling the vehicle.

What if the human is not ready to take over again?

How many times are humans not ready to drive their cars properly now?

If the human is required to sit at the ready on guard to take over when the automated system suffers a sensory failure, why have the automated system in the first place?

Because it has the capacity to be right (read: safer) more often, on average, than a human driver, on average.

 

[RC45]

Whatever tone you want to assign to my posts, feel free. I do find it strange you think the AUP is below you, though: so once more, with feeling.



Video of a car on a track. Not the road-going scenario (nor speed) your hypothetical situation originally alluded to. Not that that matters; just like you've done numerous times in this thread, if evidence is provided that doesn't align with your example, you'll just move the goalposts to another equally-unlikely hypothetical.



You're... you're serious?

Oh.

Automated systems are intended to minimize human error, yes. Glad you're finally on board. Why pass control back to the human if there's a sensory failure? Because the other option is to have no party controlling the vehicle.



How many times are humans not ready to drive their cars properly now?



Because it has the capacity to be right (read: safer) more often, on average, than a human driver, on average.

Sort of like the Canadian free health care system? Mostly right some of the time?

BTW, feel free to go ahead and prove any of my positions incorrect.

 

[SlipZtrEm]

Sort of like the Canadian free health care system? Mostly right some of the time?

That's actually an apt description of the current motorist landscape. 👍

BTW, feel free to go ahead and prove any of my positions incorrect.

That's not how it works. The burden of proof falls to those making the claim.

 

[RC45]

That's actually an apt description of the current motorist landscape. 👍



That's not how it works. The burden of proof falls to those making the claim.

Well you seem to be claiming my positions are incorrect. Where is your proof?

That's not how it works? As if I give a hoot how 'it works' LOL.

 

[Tornado]

So the entire automated freeway commuter infrastructure now relies on those pesky humans again.

Not entirely sure what "automated freeway commuter infrastructure" you're attempting to claim I'm talking about, but yeah. When those sensors required to run it fail, it probably would be a pretty good idea to alert the driver and have him respond to the failure, or have the car itself respond as necessary (like, perhaps, pulling over onto the side of the road). Just like when:

3 of the 4 scenarios will induce a reduced power/shutdown mode - the 1 of the 4 will shutdown one of the aid systems and if another system should suffer a sensory input failure will shut the car down.

So what you're saying is that the limp home mode first introduced on higher-end 1990s cars has since proliferated down the line to cheaper automobiles, and in the process included more and more passive safety monitoring systems.




Which apparently supports your point that such a thing wouldn't happen with automated cars?

With the automated car we are no longer just talking about passive safety response systems or engine management systems - we are talking about proximity sensors, distance sensors, closing rate sensors, speed sensors - systems used to place the car in a known position on the ground to help navigate obstacles etc.

Yes. Pretty complex system, making it all the more curious your insistence that any one part of it going tits up would cause all of the rest of them to fail so the car was "deaf, dumb and blind"

So, what happens when 1 of those sensors fail? You don't think it will 'shut down entire subsystems'?

No. That would be a spectacularly badly designed system if, say, the proximity sensors on an automated car failed and caused the automated mode to instantly fail and drive off the road; since in your example the proximity sensors took all of those unrelated systems down with it.

 

[McLaren]

So you propose the automated car to eliminate human error, but when the automated system suffers a sensory failure your pass control to the human?

I didn't propose the automated car to eliminate anything. I've merely called all your claims against not needing it incorrect because you have yet to provide an actual scenario where the computer did stand at fault.

As for your second statement, try not to be so vague. If a person is driving along & a system fails without any excessive load, then yes, it is a system failure. Your scenarios are not that. You dreamed up scenarios where a human made the error & made it out the computer is at fault for not saving them. You accuse me of thinking a computer is perfect in all aspects, yet in all of your examples, you expect it to be perfect at saving drivers from their own mistakes.

Why bother arresting control from the human in the first place then?

That's not what anyone has said. But, you've proven you're pretty good at not reading anything.

Forget unfortunate predicament - what about under normal circumstances if the automated car suffers a sensory failure? After all, even at norm al speeds the car is deaf, dumb and blind if it suffers a sensory input failure and all the logic in the world cannot help an input less computer.

In this situations, yes, the computer is at fault, but no one has said otherwise. None of your previous claims have been proposed this way, however.

You do understand how input driven systems operated right?

You've done a pretty piss poor job showing you have any understanding of them, yourself, so you're in no position to question me or anyone else.

That's not how it works? As if I give a hoot how 'it works' LOL

 

[niky]

Took Google .5s to come up with this video demonstration of an example of Ice Mode intervention causing issues.

Ice mode doesn't work properly in the absence of ice.

Who knew?

Also.. poor driving technique. If you're jumping hard on the brakes when the front end is unloaded, you can freak out non-ice mode ABS systems, too. And you can cause non-ABS systems to lock and send you off the road.

The problem here is not too much automation. It's that the manufacturer has layered two different systems on top of each other without having them communicate.

In this case, ice mode over-rides ABS without letting it do its job... when ABS already does what ice mode is supposed to be doing, in the first place.

This is the opposite of how you'd implement a driverless car, where a single system incorporates multiple inputs and outputs to increase safety.

-

Driverless cars are already driving our roads. Imperfectly. But they're there. Eventually, they'll be able to match the average driver, only with faster reflexes and the ability to stop quicker in an emergency.


-

Expense is one thing, but considering Ford already has automated parking (not very great, but still automated) and low-speed emergency brake assist on the Focus at less than $25k, it's not far-fetched to think semi-automated driving will be coming to mid-priced cars in the medium term future.

 

[RC45]

Ice mode doesn't work properly in the absence of ice.

Who knew?

Also.. poor driving technique. If you're jumping hard on the brakes when the front end is unloaded, you can freak out non-ice mode ABS systems, too. And you can cause non-ABS systems to lock and send you off the road.

The problem here is not too much automation. It's that the manufacturer has layered two different systems on top of each other without having them communicate.

In this case, ice mode over-rides ABS without letting it do its job... when ABS already does what ice mode is supposed to be doing, in the first place.

This is the opposite of how you'd implement a driverless car, where a single system incorporates multiple inputs and outputs to increase safety.

-

Driverless cars are already driving our roads. Imperfectly. But they're there. Eventually, they'll be able to match the average driver, only with faster reflexes and the ability to stop quicker in an emergency.


-

Expense is one thing, but considering Ford already has automated parking (not very great, but still automated) and low-speed emergency brake assist on the Focus at less than $25k, it's not far-fetched to think semi-automated driving will be coming to mid-priced cars in the medium term future.

Wow - so much wrong in a single post.

'Ice Mode' is simply one of the many ABS/AH/TCS software subroutines. It is not a separate system. So the process of the Ice Mode routine is ABS in operation - it is modulating the brakes in this case providing sub-par braking forces. ABS equipped cars are notorious for being slower to stop on ice than non-ABS equipped cars.

 

[niky]

Yes and no.

Reading through, Ice mode on the Corvette seems to be a separate software subroutine used to over-ride regular ABS, not allowing it to do its job properly. Because ABS cycles too fast for icy roads. Especially newer ABS systems, some of which cycle insanely fast. It's something that can be avoided by adopting full 4-channel ABS sensors and more sensitive yaw sensors to the car.

Poor communication... the car cannot see the condition of the road and decides it must be on ice, not accepting any feedback to the contrary. Kind of stupid. This is the kind of thing you use EBD for. Have EBD reduce force to the rear axle and allow full force to the front brakes. Allow the car to wash wide a little bit on ice relying on ABS to keep things tidy.

Or you could just threshold brake like you do with a non-ABS car. Stab the brakes on those like you do in an ABS car and you'll be off nearly just as quick. With nasty flat spots, to boot.

This is not something that would happen on a fully automated car. Automatic braking algorithms typically don't stab the brakes like a moose in heat, causing the tires to instantly lock-up, cycling ABS.

 

[Joey D]

I don't think I've ever really heard of "ice mode" being too much of an issue for the average driver, which cars now are built for. I think the only real complaints I've heard about it is when a car is used in a way the OEM never intended, such as SCCA stock racing or putting a non-OEM brake setup on the car. Sure there are going to be people who aren't the average driver, like those who visit automotive related forums, who will complain about it but that's the minority by far.

Also I can't ever recall a time where any of my cars caused instant death on ice. I'm sure living in Texas you probably have a superior idea of how to drive in less than favorable conditions over someone like me who lives in Western Michigan, one of US's worst places for winter, though.

 

[niky]

I've never seen it myself during our braking tests. Then again, when we do our braking tests, they're typically in a straight line. I'm thinking we should add our own version of the moose test, but those are very tricky and dangerous to set up right.

Over the past few years, I've actually seen no advantage in having no ABS on wet or dusty surfaces in terms of braking distances. Once the tires start sliding, ABS gathers it up faster than trying to manually pulse the brakes. Threshold braking might just be faster, but that requires you to correctly identify how much braking force you can use on the surface you're on every single time. On the race track, doing multiple laps of the same circuit, with good tires, this may be possible. On the open road, in changing and varied conditions... it's a whole lot harder.

EDIT: On my old Sentra, however, the crude 90's era ABS was so bad that it would cut braking almost completely every time the road was even remotely rough... even at very low speeds. But time moves on, and ABS is getting so good that you have to basically look for high-speed, high-performance driving examples to find any instance of this happening.

 

[MatskiMonk]

ABS really saved my ass last week, and I don't mind admitting it - I'm sure the guy in the 5 series in front, and the Golf behind, were both equally happy with the performance of their systems too as it avoided at least a 4 car shunt in the outside lane of the M42 last week - I'd hate to be the driving enthusiast purist that got that one wrong, and having to explain to multiple other parties that it's because I shun modern technology! I don't understand why people aren't happier to embrace new technology, it's there for our benefit.

 

[Wolfe]

The Legacy is the first car with ABS that I've driven more than occasionally. Whether it's because of '90s crudeness or not, it once tried to pitch me off of a muddy tree-lined drive because brake lock was verboten and it was more than happy to allow gravity to take over instead of stopping the car. Being inexperienced with ABS, it was shocking having a computer nearly second-guess me into a ditch. I only stopped when it occurred to me to pull the handbrake. After some experimentation on the icy driveways I navigate for my job, it was more than obvious that ABS was a liability. Now I drive with the fuse unplugged most of the time.

Similar to what I said in my first post, I would much rather accept full responsibility for an accident than be blamed for something that may have only occurred because ABS wouldn't permit the braking force I needed.

 

[MatskiMonk]

The Legacy is the first car with ABS that I've driven more than occasionally. Whether it's because of '90s crudeness or not, it once tried to pitch me off of a muddy tree-lined drive because brake lock was verboten and it was more than happy to allow gravity to take over instead of stopping the car. Being inexperienced with ABS, it was shocking having a computer nearly second-guess me into a ditch. I only stopped when it occurred to me to pull the handbrake. After some experimentation on the icy driveways I navigate for my job, it was more than obvious that ABS was a liability. Now I drive with the fuse unplugged most of the time.

Similar to what I said in my first post, I would much rather accept full responsibility for an accident than be blamed for something that may have only occurred because ABS wouldn't permit the braking force I needed.

I had a 1983 BMW 6 series with the clunkiest ABS in the world, it did just as good a job stopping me on ice as I could have done with my right foot*. I've also attempted to use the handbrake on sheet ice, and unsurprisingly, it did nothing.

Either way, as a driver you have to accept responsibilty. I found myself more likely to rely on ABS in sudden emergency braking situations, when I'm sure most drivers simply cannot react as well as ABS. In situations where the road is Icy, I would think most people would have adapted their driving style and eased off a bit.

*actually left foot since it was an Auto

 

[TenEightyOne]

'Ice Mode' is simply one of the many ABS/AH/TCS software subroutines. It is not a separate system. So the process of the Ice Mode routine is ABS in operation - it is modulating the brakes in this case providing sub-par braking forces. ABS equipped cars are notorious for being slower to stop on ice than non-ABS equipped cars.

I learnt to drive in the North East of England and the lessons included switching ABS off due ta tha snar leek. In those days (as mentioned by other posters) it was pretty clunky and brutal... and it was utterly ineffective on ice. If anything it could unbalance the car more than being in a "controlled" skid.

 

[niky]

It must be said... driverless cars will not need ABS. Instead, the systems will be able to modulate the brakes themselves and should be able to perform threshold braking, probably in conjunction with wheel speed sensors with finer resolution than current ABS sensors. (Which are again, better than old ones... though many cars still use crap ABS... or whatever bare minimum is required to claim having ABS)

 

[AudiMan2011]

Thread related:

The autonomous RS7 has been upgraded and was able to shed 400kg in weight. It recently lapped Sonoma in just over 2 minutes. Though the claim that the lap is "better than those of sports car drivers" is really hard to believe.

http://www.carscoops.com/2015/07/improved-audi-rs7-piloted-driving.html

 

[LivingWithGames]

I'm not fond of autonomous cars either, @Slash . I'd never get comfortable with a car living on it's own.

But while I have my doubts, and somehow, trust or believeness inthis, I can almost see the sight, if no human beats a car driving itself, then we can say that the line....Crap, I forgot the quote or line but something related that, the machine and the human body being one, will be finally be erased.

This is at same time impressive and scary. I can't stand driverless cars at all.

 

[ScottPuss20]

Sitting in front of the steering wheel and not using it makes no sense to me at all. I hope that autonomous cars don't take over our roads, I wouldn't feel content with that. It's technology for the sake of technology, there's no point investing tonnes of money into something that isn't really needed.

 

[Joel]

Sitting in front of the steering wheel and not using it makes no sense to me at all. I hope that autonomous cars don't take over our roads, I wouldn't feel content with that. It's technology for the sake of technology, there's no point investing tonnes of money into something that isn't really needed.

There were over 30,000 road fatalities last year in the US alone. You don't think a technology which could eliminate the large portion of those accidents caused by human error is needed?

You don't think this technology is important to people? Have you ever had to commute an hour or more a day? Think about how much it would improve people's quality of life to sit and relax for an hour on the way home instead of driving in heavy traffic every day.

 

[Pseudopod]

Not that I've driven a lot of different cars on snow, but I haven't had much trouble with my 90s audi's ABS system. Naturally, my car doesn't stop very quickly on sheet ice, but I doubt it would without ABS either. That's probably got more to do with me not using studded winter tires.

 

[McLaren]

There were over 30,000 road fatalities last year in the US alone. You don't think a technology which could eliminate the large portion of those accidents caused by human error is needed?

You don't think this technology is important to people? Have you ever had to commute an hour or more a day? Think about how much it would improve people's quality of life to sit and relax for an hour on the way home instead of driving in heavy traffic every day.

Have to agree. I don't think this technology's purpose is to completely remove driving from humans (yet), but ease the process.

My only fear would be drunks finding it acceptable to go home in the car because it drives itself.

 

[MatskiMonk]

My only fear would be drunks finding it acceptable to go home in the car because it drives itself.

Still better than them finding it acceptable to put the keys in ignition and the try and operate the wheel, pedals and sticks themselves though surely...

 

[AudiMan2011]

I'm a bit skeptical about autonomous cars due to the faith and trust in the technology that's needed, particularly in this infancy stage where it's starting to roll out on production models, but as time goes on and as the tech improves, we will likely become more accepting of autonomous driving as it might become a new norm. It's the next leap in automotive technology and I for one, welcome it for the time being.

Of course there will still be a hardcore minority of people out there screaming "muh freedoms" as autonomous cars become more widespread..

 

[McLaren]

Still better than them finding it acceptable to put the keys in ignition and the try and operate the wheel, pedals and sticks themselves though surely...

It's not acceptable, either way.

 

[Joel]

Have to agree. I don't think this technology's purpose is to completely remove driving from humans (yet), but ease the process.

My only fear would be drunks finding it acceptable to go home in the car because it drives itself.

What's interesting about the Google prototypes and other self driving cars that are being tested is they're already functional in a way that minimizes the downsides and maximizes the quality of life improvements from the cars. People seem to be more afraid of them driving on the highway erratically and less afraid of them around town, when the tech right now is actually the other way around. Relatively speaking it's almost trivial to have these things on the highway, all the cars are going the same direction and a similar speed, so it's just a matter of sensors around the car and regulating the speed and steering to keep a safe distance from other cars.

The challenges right now are getting the cars to interact with city traffic and the irregular patterns. Things like a delivery truck parked in a lane, a police officer directing traffic, or a pedestrian/cyclist motioning to a car to go. It's hard for a computer to interpret those symbols because they're kinda ambiguous and hard to program for. We're a long way away from fully automated delivery trucks that can navigate downtown traffic, but Google's prototypes are already really capable on the highway. I agree though, we're a long way away from having cars that can take over for drunk drivers entirely, I'd imagine the laws will require you to be sober and awake as they'll all still have manual override for years.