It kind of seems like a “map this road” feature would be relatively easy to make, and would, in fact, be a critical part of making these things work to begin with. Say my car finds itself on a road like enipla’s driveway for the first time, and that the first time it encounters that road it’s not buried under a foot of snow. I hit the “map this road” button, and it memorizes potholes and exact coordinates of the turns and trees and whatever, and then when it is under a foot of snow, it still knows those coordinates. And presumably, it would then upload this info so other cars can access it.
If we’re talking public roads, I would expect these cars to be constantly transmitting data about the surfaces and obstacles and whatnot, so that if my car sees a new pothole in the road, other cars know to avoid it too. And the city knows about it and fixes it.
A good idea in principle but how would the car navigate? GPS is not accurate enough to reliably stay in a lane, let alone avoid pot holes. Autonomous car tech is based on “seeing” the road and obstacles directly a bit like we do. You can make GPS more accurate by having a ground station that knows its position and knows what a GPS says its position is and broadcasts a correction factor to near by receivers, but that would require significant infrastructure.
Sounds like it could be pronunciation problem. Does the pronunciation of high end sound similar to lived or dead? If the pronunciation does not have hard ed sound I may be having a pronunciation problem and it sounds like in to me.
I know I have pronunciation problem some times with in and im like imported vs inported sounds same to me.
Driving an Interstate is easy - that will be .05% of the logic/infrastructure needed.
Standard ‘grid’ streets - easy.
Keep going in that “easy” - “hard” path and you end up with snow-covered ruts in a gravel road that turned to mud a year ago.
It’s going to be a loooong time before that one is fixed.
Infrastructure requirements:
Traffic control must send machine-readable instructions - a light bulb or metal sign is worthless.
If we expect the car to figure things out, we will need WAAS (the GPS correction devices) simply EVERYWHERE.
When you finish the list of infrastructure updates, you are going to be hit with the obvious: the places which require the most cost-intensive upgrades are the places that get the least money - those washboard gravel roads, the culverts which flood for a week every spring, the remaining one-lane, 5-Ton limit iron bridges - and there ain’t going to be nobody willing to spend that kind of money on those back-woods County Roads.
So your shiny new car has large holes in its map/algorithms - “I can’t go there, Dave”, throws ups its hands and dies if driven into one of these black holes.
Yes, there will be manually operated vehicles for another 100 years.
I love driving! But only when moving. Being stuck in traffic is pure hell.
While I’m not a billionaire, I’ve had a driver for the five years I lived in China. The only real convenience was for being able to imbibe to a degree that I otherwise wouldn’t be able to. Well, and not having to find parking on those occasions that I’d venture into the city.
Autonomous cars would solve all of the problems we had with drivers, though, like trying to ensure his work-life balance, sick days, fake fuel receipts, unexplained extra mileage. Having a human driver means having less freedom.
It will be like the high-end airlines are now - there’s an “auto-pilot” that can do all the driving on a typical trip, but still an ability to drive manually for situations not anticipated by software designers or an actual human may be preferable.
For example - day-to-day driving is handled by the car, but a driver encounters a blocked road (accident, landslide, volcanic eruption, whatever) and the authorities want to route the blocked traffic off the road, a short way through a field, and back onto a road through a street light that is stuck on red. The car’s software, which is undoubtedly programmed to obey the law, at best is going to require someone constantly saying “do this anyway” and at worst simply won’t cooperate. A human, however, can recognize this is an unanticipated situation and follow the directions of a cop directing traffic and ignore the ordinary rules (stay on the road, don’t go through a red light) for the duration of the circumstance.
Considering I’m driving a 2007 bottom end Legacy that has all those features, I’d say you’re off track. Did the car originally come with them? No. I did, however replace the OEM radio with a Bluetooth enabled unit for $250. I now have every single item on your list courtesy of Google and my Samsung S4 Mini for a total outlay of less than $650. Throw in a few monitors for the back seat and you’re still under $1000.
That pretty much covers it LSL. As it is both my Wife and I often turn off the VDC/traction control (2016 Subaru Outback, 2004 Nissan Pathfinder). When the going gets tough (deep and steep) it causes problems. I guess manufacturers recognize this, or their wouldn’t be an off switch. Sometimes you just have to intervene, the car does not always no what’s best.
In many dialects, end does not have a d sound at the end, but the vowels in end and in are different.
In any case, the meaning of the two words is so different that I really have problems understanding how can anybody confuse them.
Also, pronunciation problems do not explain why did you lose the “a” before pronunciation. Your mother tongue apparently uses less articles than English does.
Sounds about right to me, probably on the earlier side, but I’m not 100% sure. My basic 2004 Mazda didn’t have it. My 2014 did. So somewhere in between.
For a stretch of road like a driveway, a computer can accurately map out the exact location of everything within sight by using a few fixed points of reference, the house, a tree, a shed, etc.
The computer won’t forget where the driveway is, even under a foot of snow.
Many of the ‘black hole’ areas can be mapped out one time and that information uploaded to cars that need it, with the information refreshed every single time an automated car goes through that area.
In bad conditions, will the computer be smart enough to take the driveway from a different route? Say going down the road, turning around and making a run at it so you can get more speed up? If it doesn’t make it up the first time, how will it decide if it should give it another shot? Perhaps with a little more speed or 6"s to the left or right. If it can’t see beyond a snowdrift, will it plow through? How easy will I be able to navigate it to anywhere in my yard or off road in places that it has never been?
I’m constantly baffled by the people who say “Sure, a computer might be better than a human, but…”, and then proceed from the assumption that the human is better. How will the computerized car park in the middle of a grassy field for a special event? Why, it’ll follow the guy wearing the orange vest and the batons gesturing you into the appropriate row, of course. The same way you do it.
EDIT:
Case in point from enipla. Can a human do those things? Then a computer can do them the same way.
And if there is no human in an orange vest and batons?
As I’ve said upthread, at least currently, systems that are supposed to help, like VDC have to be turned off manually to handle some situations. The computer thinks it’s helping by cutting engine power (for instance) but only manages to make things worse.
I’ll give people a pass on not understanding this as most don’t live in an area that gets from 20-30 feet of snow a year.
These conversations sometimes get stuck on the difference between what self-driving cars can do in theory and what self-driving cars can be reasonably expected to do in the near future.
There’s really no known limit to what AI can accomplish. There’s no special realm of human thinking that can’t be emulated by software. In theory. That doesn’t mean that software is going to be sophisticated enough in the next 5 years that it will deal with all the things human drivers are expected to deal with on a regular basis.
So I have no problem believing that a car could handle a grassy-field-parking-lot just fine, someday. I do have a problem believing that a car is able to do so today, or even in the next few years. When you add in all the big, little, and medium problems that self-driving cars will have to handle I think it is ludicrous to think they will do so in the next five years. The next fifteen, maybe, if we’re lucky. (My best guess is 30+ years before 99%-autonomous vehicles are mainstream, although “mostly”-autonomous vehicles will be mainstream long before then.) They will continue to improve over time as engineers chip away at the problem of teaching a car to drive itself, but there are a lot of problems and they aren’t all as easy as maintaining speed and staying between the lines on a well-marked highway in good weather.
Open highways, in good weather, we are nearly there. As long as a human is in the drivers seat. Though I worry that it will be easier, and easier to become complacent, when you really are still in charge.
My Wife’s new 2016 Subaru has all the bells and whistles. Never tried it my self, but I think it would do OK on it’s own on I70 in Kansas for a bit.
Anyway, said super smart car discharged it’s battery. Dead as a doornail. Had to get it towed (under warrantee, so what the heck). Well, the electronic parking brake was engaged. Can’t disengage E brake without power. Can’t get it on the flatbed with brakes engaged. Oops. A simple (if pain in the ass) thing that was corrected by putting a battery charger on it for about 30 minutes.
Not a confidence booster that the engineers did not think of such a simple issue.
A computer is better than a human at some things, but a human is better than a computer at some other things. With current tech a computer is not better than a human at all aspects of driving but it is better at some aspects of driving. People are just pointing out the areas that will be a challenge for the development of fully autonomous cars.
As for your guy with batons example, you say it like it is the easiest thing in the world for this to be programmed, but is it? I think that kind of functionality will present enormous challenges to the humans who are creating these cars. It may never work reliably. It may require an entirely different and more computer appropriate method such as parking areas being setup with portable reference towers and cars entering the parking area are given a local coordinate to park at and a route to get there. The point is, this is not trivial, and may not be happening in the next few years.
As said by BlackKnight, there may be a time in the future when computers are better than humans at all aspects of driving. The question is, when will that be? 5 years?10 years? 50 years? Never?
Until a computer can do everything in a car that a human can, there won’t be any widespread use of cars without manual controls. As for computer controlled cars with manual controls, they’ve been around for decades and we are seeing an increasing number of things that the computer can do, e.g., lane following, parallel parking, adaptive cruise control. Just recently there has been a huge increase in what the computer can do. As we all know, test cars are now able to autonomously do most of the driving that a human can do. But they’re still not doing everything, there are still challenges to be overcome.
Humans can’t do those things without input from humans?
And sure, there are some things that computerized cars can’t handle right now. Nobody’s claiming the technology is currently mature, or even really ready for general adoption. But there’s no reason they can’t be. Interpreting human gestures, for example (like those of the guy with the orange vest and the batons) is something that mass-market consumer-grade video game systems have been doing for years now. There’s no reason you couldn’t give a computer driver the same capabilities as a Kinect.