|
|
This topic comprises 3 pages: 1 2 3
|
|
Author
|
Topic: Reclining seat death
|
Marcel Birgelen
Film God
Posts: 3357
From: Maastricht, Limburg, Netherlands
Registered: Feb 2012
|
posted 03-24-2018 03:14 PM
quote: Leo Enticknap
In a vehicle with a mode selector that is not a physical lever operating switches and relays, this is at least theoretically possible. Visiting relatives in England a couple of years ago, Hertz gave me an "upgrade" to a high end Merc at Heathrow (translation: they foisted it on me because its fuel economy is so sh!tty that no-one else would take it), in which the transmission mode (P/R/N/D) was selected from a touchscreen on the dash.
I've never seen a Merc with only a touch-screen drive mode selector, not denying it doesn't exist but it sounds rather awkward. The newer Mercs with AT do have mode selector on a stick behind the steering wheel, the stick in the center console has been gone for a few years now.
quote: Leo Enticknap
So if a software bug decided that it didn't want to let me shift from D to N, and then the handbrake cable snapped and all the hydraulic fluid leaked out of the master cylinder, the result could have been quite an impressive bang-ding-ow.
This sounds more like a bad scene from Final Destination 29, than a movie adaptation of "The Machine Stops". ![[Wink]](wink.gif)
Yes, if multiple things fail all at the same time, things are looking a bit gloomy for you. Stuff should be designed that, as long as it is properly maintained, it should simply not happen...
Then again, I was somewhat in disbelief when, after a colleague had the windshield of his BMW replaced with a wrong model, the on-board computer would freak out at random intervals and block random functions. At one time, we couldn't get out of the car the normal way, because the doors refused to open, the only way to leave the car was via the windows... This got me scared a little, because the last thing you expect to block in a car is the functionality to open the doors from the inside...
quote: Mike Blakesley
The thing I don't understand about self-driving cars is, why are they trying to design them without a steering wheel? There's bound to be situations where you would NEED to maneuver the car manually...maybe on a road with a weird blockage like a fallen tree branch, or a huge pothole, etc.
From what I know, only Waymo is currently considering removing the classical steering wheel. It doesn't mean it will not return another means of manual steering, like a joystick for example.
Removing manual steering entirely, would require cars so smart, they can handle all unforseen circumstances themselves. This should include stuff choosing a correct parking spot at your destination, which is a pretty hard task to achieve.
quote: Mike Blakesley
Also what about snow-covered roads, gravel roads, roads without stripes, blind corners, and so on? I agree the self-drivers are the future, but taking out all the controls seems kind of stupid.
In theory, a self-driving car could even perform better on those condition than a human. Many of the current prototypes have much ore inputs than any normal human has and could process. One of those is lidar, which is laser based 3D mapping of the environment. The car could be better equipped in mapping what surfaces are drive-able than a human driver.
Also, we could optimize our roads for self-driving vehicles by putting stuff like RFID beacons into the road itself. That way, the car could determine exactly where the road is, even if it is covered in feets of snow.
| IP: Logged
|
|
|
|
Marcel Birgelen
Film God
Posts: 3357
From: Maastricht, Limburg, Netherlands
Registered: Feb 2012
|
posted 03-25-2018 03:11 AM
quote: Mike Blakesley
I could be wrong but I seem to remember an article about the Google car that had no wheel but who knows.
Waymo is the current brand the Google Car stuff was moved to under the "Alphabet" holding, so we're talking about the same cars.
quote: Mike Blakesley
As for the amazing ice and snow stuff, I'll believe it when I see it working. There are so many more variables to figure out when driving in the winter. The vision part is the easy part. What snow does to the roads and the vehicle is the tricky part.
Well, think about it. Even now we have cars that help people to stay on the road in bad weather conditions. Stuff like ABS, ESC/ESP has saved many lives over the years and it is generally improving.
A computer can evaluate hundreds of thousands of inputs a second and take split-second actions on any of these. That's beyond anything a human can do.
quote: Mike Blakesley
I'm not saying all those problems and others won't be tackled and taken care of, but the fact is that most of the tests so far have taken place in large cities with good street markings in nice (maybe rainy) weather. Bring the Googlemobile to Forsyth, Montana and lets see how it does. It just seems that Ma Nature, crafty old broad that she is, every time you think you have her tamed, she throws another icicle at you.
Waymo claims to have tested their cars in the most adverse weather conditions. Now, I don't know if that includes totally snowed over roads, but they have driven a hell of a lot of miles already.
Still, the biggest challenge doesn't even seem to be Mother Nature, but a factor that's far more incalculable and that's the other humans that share the road with those self-driving vehicles.
Where humans often use intuition to avoid potential trouble, machines have no such thing.
Consider a multi-lane freeway. To effectively drive on such a road, sharing a lot of traffic with you, you depend a lot on others sticking mostly to the rules. For example, you expect people to stay in their lane, if there's no reason to change.
Now, we humans often can see if another driver is going to change the lane and we can anticipate on it by e.g. adjusting our own speed. We can see he's been stuck behind this big truck for a while now, has already made a few tentative moves and he's going to pass it on the left any time now, on the same lane you're driving. Since this person is somewhat of an *sshole, he's not going to respect the considerable difference in speed between you and him and you're ready to give your brakes a good workout.
It's not an entirely unsolvable problem for a computer, but it is far more difficult. Whether or not to reduce speed in such a situation is far more difficult for a computer to decide, because there is no such thing as intuition. So, a computer will more often choose for the safer alternative of reducing speed, even if you would not do so yourself, based solely on your intuition.
You see this already in semi-automatic drive modes from manufacturers like Tesla. If you're on a multi-lane road and it needs to overtake some obstacles, you'll see it will often reduce speed before it does so. Simply, because their algorithm assesses the risk differently than you do.
| IP: Logged
|
|
Mike Blakesley
Film God
Posts: 12767
From: Forsyth, Montana
Registered: Jun 99
|
posted 03-25-2018 01:13 PM
quote: Marcel Birgelen
Waymo claims to have tested their cars in the most adverse weather conditions. Now, I don't know if that includes totally snowed over roads, but they have driven a hell of a lot of miles already.
I appreciate that, and I agree with your points, especially about intuition. I feel like this is something that can indeed be figured out, but it will take an amazing amount of onboard computer power and analytics to do it.
The most unfortunate fact about all this is that sometimes computerized stuff just doesn't freaking WORK. For no reason at all.
I have a fairly new (2016) Ford pickup which has "backup assist," which includes the camera, sensors, radar and whatever else to detect obstacles. If it's cold outside, sometimes -- NOT ALWAYS, but sometimes -- that particular system just refuses to work. It'll just emit an annoying beeeeeeeep until I either turn it off, or stop backing up. Then the next day, under the exact same conditions, without having cleaned the sensors and whatnot, the same system will work perfectly. I hear other similar stories about electronic "systems" frequently from other drivers during my work days (at the parts store).
It just seems like if carmakers could get these basic, comparatively simple systems to work 100% of the time first, they'd go further toward getting the confidence of the public in self-driving cars.
| IP: Logged
|
|
Leo Enticknap
Film God
Posts: 7474
From: Loma Linda, CA
Registered: Jul 2000
|
posted 03-28-2018 08:20 AM
quote: Mike Blakesley
If it's cold outside, sometimes -- NOT ALWAYS, but sometimes -- that particular system just refuses to work. It'll just emit an annoying beeeeeeeep until I either turn it off, or stop backing up.
This illustrates why automated systems that are designed to be relied on to carry out safety-critical functions that require intelligence-based decision making, are an order of magnitude more difficult and expensive to create than ones that are either not safety critical, or don't have to be totally relied on (in other words, the system is designed with the expectation that a human can and will take over if necessary, and will also be able to judge if and when it becomes necessary; or that the consequences of the system failing are non-serious enough for that risk to be considered acceptable).
In the case of your backup assist system, I'd guess that the designers went so far and no further with the hardware and software development, because they were inventing a labor-saving aid, not a safety critical automation system, and therefore they didn't have the budget needed to create a system that was 99.999% guaranteed to work in all weather and all traffic situations. Say, 80%, was as close as they were aiming for, and expected to achieve, because you can always back the vehicle up yourself if needs be.
However, when you're in a situation in which the car's owner expects to get in, tell it where to go, have a nap and then wake up at their destination, the engineering challenge suddenly becomes a very different one.
The really scary aspect of all this to me is when the boundary between those types of system (ones that have to be relied on to operate autonomously, and those that don't) get blurred. Air France Flight 447 is a particularly scary example: it appeared that none of the (human) pilots had been properly trained to understand how the autopilot acts when it no longer knows the plane's airspeed reliably, and possibly didn't know how to fly the plane out of trouble manually in zero visibility (total darkness and complete spatial disorientation outside) and without knowing their airspeed reliably. Their training and expectation was quite simply that what happened would never happen, and therefore that it was unnecessary to know how to deal with it if it did.
| IP: Logged
|
|
Dave Macaulay
Film God
Posts: 2321
From: Toronto, Canada
Registered: Apr 2001
|
posted 03-28-2018 10:43 AM
General Ludd here.
There are fewer reasons than for flying cars, but - like flying cars - level 5 autonomous vehicles are not possible here.
Our landscape has been designed around cars - there are basically no places in North America where cars and roads don't dictate planning decisions.
Cities and towns in North America are designed for cars and people to coexist. Level 5 autonomous vehicles have unavoidable problems with driving in cities although they could potentially work on superhighways (in most weather) but what happens when they need to go off the "interstate" into town?
They will always have to avoid hitting people. People are both stupid and smart. In a 100% autonomous vehicle city or town people are smart enough to very quickly realize that they can walk across any street any time, that the cars have to stop to avoid hurting them. And stupid enough to do it all the time. Traffic would be paralyzed.
The only answer is to have all roads sealed off from pedestrians. That's already done in some parts of the world for major city arterials but I mean every road in every urban/suburban setting. The cost of redesigning every city is insane.
Not. going. to. happen.
There is one overpowering reason that even level 4 autonomous vehicles are unlikely - a driver has to be driving to pay attention to the driving environment. Expecting a person who - with no reason to watch the road - is reading, sleeping, screwing... whatever, to take command immediately when the autopilot sounds an alarm due to some situation it can't handle - likely something dire - ridiculous. Check the video from the fatal collision (this is a level 4 vehicle) : this person is being paid to stay alert and be ready to take control of the vehicle instantly, that's his only job, and he knows that cameras are continuously recording him so he'd better stay sharp because jobs that easy are hard to find.
And? He fails, looks away from the road for long periods, but happens to look up a moment before impact - probably because some alarm has sounded when the autonomous system realized the collision was imminent (there was no sound on what I saw). Yeah. Great plan.
| IP: Logged
|
|
|
|
|
|
|
|
Mike Blakesley
Film God
Posts: 12767
From: Forsyth, Montana
Registered: Jun 99
|
posted 03-28-2018 06:15 PM
To be fair, it's pretty hard to expect a person to sit there, doing absolutely nothing, but still being responsible for the entire operation of the car. After about 10 minutes of being completely bored, of COURSE the typical person's going to start looking at their phone (or out the side window, or at a newspaper, or at their companions). I enjoy driving, but I would hate driving under those circumstances. What a drag it would be.
Until a self-driver simply never NEEDS intervention from a human while on a roadway, mass acceptance will still be uneasy at best (well, until everybody now over 30 is dead).
| IP: Logged
|
|
|
|
Leo Enticknap
Film God
Posts: 7474
From: Loma Linda, CA
Registered: Jul 2000
|
posted 03-30-2018 08:34 AM
For these reasons, I predict that level 4 and 5 autonomous driving will be allowed on freeways and some other roads on which non-motorized traffic is either segregated or almost non-existent (e.g. state highways across the desert), long before it's allowed on inner city and suburban streets.
In the same way that shipping containers are now moved long distances by rail, but transferred to and from a big rig trailer at depots for the short distance between the railhead and the origin and destination points, I wouldn't be surprised if something similar happens for road traffic. The vehicle would be driven by a human as far as the freeway on-ramp, or in the case of a big rig, to a lot next to the on-ramp, where the driver gets out. The human in a private car puts the car into autopilot as it enters the on-ramp, and can then relax until it's time to leave the freeway. The big rig drives itself to a lot next to the off-ramp closest to its final destination, where a human driver boards and takes it the last mile.
| IP: Logged
|
|
|
|
|
|
|
|
|
|
All times are Central (GMT -6:00)
|
This topic comprises 3 pages: 1 2 3
|
Powered by Infopop Corporation
UBB.classicTM
6.3.1.2
The Film-Tech Forums are designed for various members related to the cinema industry to express their opinions, viewpoints and testimonials on various products, services and events based upon speculation, personal knowledge and factual information through use, therefore all views represented here allow no liability upon the publishers of this web site and the owners of said views assume no liability for any ill will resulting from these postings. The posts made here are for educational as well as entertainment purposes and as such anyone viewing this portion of the website must accept these views as statements of the author of that opinion
and agrees to release the authors from any and all liability.
|
Home
Printer-friendly view of this topic