Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
Like I said, this has been "proposed" since the 50s. 60 years later, we're hearing about it again.
The difference being the technology is available to make it work and work well. For a car to operate safely it literally needs a mind of it's own, sensors and everything else are just window dressing. Technology that can make rational decisions of it's own based on ever changing variables in the environment is what will make this tech work.
The difference being the technology is available to make it work and work well. For a car to operate safely it literally needs a mind of it's own, sensors and everything else are just window dressing. Technology that can make rational decisions of it's own based on ever changing variables in the environment is what will make this tech work.
If they can make all that happen at an affordable cost is and always has been the question.
The ability to make self-driving vehicles has been around for decades. To do it economically while demonstrating its fail-safe reliability is still to be achieved.
If they can make all that happen at an affordable cost is and always has been the question.
No it is not the only question, this type of technology has only really started to develop over the last decade.
Quote:
The ability to make self-driving vehicles has been around for decades.
No it has not because the prerequisite technology to make it work in the real world has not been available. You cannot do this with sensors and instructions alone because you could never have enough instructions to meet every possible scenario. The computer driving the car needs to make rational decisions autonomously, going back to my original post it needs to be able to answer the question "What is the color of Dorothy's shoes?" The fastest computers in the world all working together on this question could not answer it a decade ago. Watson *may* be able to answer it.
There is significance difference between how a computer like Watson and these cars operate compared to super computers like Deep Blue from more than a decade ago.
As long as there is extreme environments that electrical components will be subjected to that can cause failure and the loss of human life, there will not be a high degree of confidence given to automated driving systems.
And, there is always something with the human element that will cause a life threatening situation.
Look at the incident the other day when a railroad worker parked his backhoe on the railroad tracks causing a train crash.
Everyone is now wondering what were they thinking ?
There is going to be a lot of failures along the way but at the end of the day a human would never be able to compete against a computer. You could have supercars safely traveling down the highway bumper to bumper at 200MPH.
This tech is going to raise a lot interesting moral questions, you're computer driven car is going down the road and it has choice of hitting the kid that just ran out in front of you or the telephone pole. That said most of these types accidents will be far less common with computer tech to begin with.
Cars are not even close to being fully automated right now and they have problems all the time and keep mechanics employed.
Sure they can make a car travel "safely" at 200 mph but what happens when it hits a random object in the road and busts a ball joint or tie rod or blows a tire out? What happens if a large object smashes through the windshield? The driving straight down the road is the easy part but having a fully automated car be prepared for every scenario that can pop up is the tough part.
All I can imagine for our future is everyone coasting along the highway at 55mph like sheep. Artificially low speed limits from the 70s.
THIS IS BAD. Anything done at whatever limit with no variation will be bad for us all. Will these cars even adjust speed against each other to change lanes, etc.?
And lastly, you think we bury our faces in devices now? Wait till we ride in these things.
Cars are not even close to being fully automated right now and they have problems all the time and keep mechanics employed.
It's not that hard to take a modern car (one with throttle and steering by wire like some Mercedes, with a computer controlled transmission, SCM, TCM, and active ABS [one that can increase and decrease braking force]). Hook up the ECU, "by wire" control computers, an external sensor array, some nav software, and the digital driver. There's not much needed to take a modern car from requiring a driver to being driven by the onboards. Most of it is just plugging in to the current electronics of modern cars.
Quote:
Originally Posted by eddie1278
Sure they can make a car travel "safely" at 200 mph but what happens when it hits a random object in the road and busts a ball joint or tie rod or blows a tire out? What happens if a large object smashes through the windshield? The driving straight down the road is the easy part but having a fully automated car be prepared for every scenario that can pop up is the tough part.
What happens when a human driver hits a random object in the road at 200mph? Can you be sure the automated driver would hit the random object? It's detection of road debris may well be significantly better than a human, so no impact no problem. If it impacts and there's an issue, the routines for most brake/chassis/suspension failures are pretty straight forward. You may consider hitting an object and the tie rod failing a different scenario to a blow out, but if the actual effect is that the car has an extreme pull to one side and reduced steering response, does it matter what the cause it? Would you try to stop and pull over or just stop in both cases? How do you adjust your steering, brake and throttle responses? Or at least stop (what changes to your braking, steering and throttle responses)? What correction routines do you use, how many maybe half a dozen or so ten at most. What's your reaction time to the failure (because reaction time is often most important) bet it isn't as fast as a computer.
Here's the thing, I'll bet you dollars to donuts that if you go through a bunch of scenario's of car failures (which is how you probably file your correction procedures), you'll find that for all there are thousands of failure scenario's you can think of but there are very limited correction procedures and how can there be thousands of correction procedures? When driving you're only really controlling the rate of increase of one dimension, and your position in the other dimension. You only have three main inputs too brake, throttle and steering, so that's really 3 variables that manage the rate of increase of one dimension, and your position on the other. It's a pretty simple combination matrix, you could probably do it with a finite state machine, no AI required.
Now if the driverless cars are all networked, then the driverless car with the failure can send that failure to every car within range and it's intent and they can all move to avoid that issue (so less problem of someone losing control at 200mph completely unbeknownst to the 20 cars behind it and the cars to it's right and left. How many RTA's are caused because the traffic behind or to the side hit or are hit by the car with the mechanical failure?
If a large object came through the windshield at 200mph, an automated car is probably far safer than a driven car, the driven car possibly wouldn't be driven any more after that object came through the windshield. It's difficult to drive without a head, even if you have a head your passenger may not, and even if you and the passenger (or there is no passenger) have heads you have significantly reduced visibility forwards (and may miss seeing the object that blows a tire or tie rod ).
It's not that hard to take a modern car (one with throttle and steering by wire like some Mercedes, with a computer controlled transmission, SCM, TCM, and active ABS [one that can increase and decrease braking force]). Hook up the ECU, "by wire" control computers, an external sensor array, some nav software, and the digital driver. There's not much needed to take a modern car from requiring a driver to being driven by the onboards. Most of it is just plugging in to the current electronics of modern cars.
What happens when a human driver hits a random object in the road at 200mph? Can you be sure the automated driver would hit the random object? It's detection of road debris may well be significantly better than a human, so no impact no problem. If it impacts and there's an issue, the routines for most brake/chassis/suspension failures are pretty straight forward. You may consider hitting an object and the tie rod failing a different scenario to a blow out, but if the actual effect is that the car has an extreme pull to one side and reduced steering response, does it matter what the cause it? Would you try to stop and pull over or just stop in both cases? How do you adjust your steering, brake and throttle responses? Or at least stop (what changes to your braking, steering and throttle responses)? What correction routines do you use, how many maybe half a dozen or so ten at most. What's your reaction time to the failure (because reaction time is often most important) bet it isn't as fast as a computer.
Here's the thing, I'll bet you dollars to donuts that if you go through a bunch of scenario's of car failures (which is how you probably file your correction procedures), you'll find that for all there are thousands of failure scenario's you can think of but there are very limited correction procedures and how can there be thousands of correction procedures? When driving you're only really controlling the rate of increase of one dimension, and your position in the other dimension. You only have three main inputs too brake, throttle and steering, so that's really 3 variables that manage the rate of increase of one dimension, and your position on the other. It's a pretty simple combination matrix, you could probably do it with a finite state machine, no AI required.
Now if the driverless cars are all networked, then the driverless car with the failure can send that failure to every car within range and it's intent and they can all move to avoid that issue (so less problem of someone losing control at 200mph completely unbeknownst to the 20 cars behind it and the cars to it's right and left. How many RTA's are caused because the traffic behind or to the side hit or are hit by the car with the mechanical failure?
If a large object came through the windshield at 200mph, an automated car is probably far safer than a driven car, the driven car possibly wouldn't be driven any more after that object came through the windshield. It's difficult to drive without a head, even if you have a head your passenger may not, and even if you and the passenger (or there is no passenger) have heads you have significantly reduced visibility forwards (and may miss seeing the object that blows a tire or tie rod ).
I like the idea of the self driving tech being used to make vehicles safer for the driver, but I don't want it to be fully self driving at all times. Keep in mind that many of today's vehicles will be on the road for more than 20 years. For this utopian dream of self driving cars to be true, all road going vehicles must be self driving. This includes tractor trailers, motorcycles, motor scooters, ambulances, police vehicles, heavy equipment (tractors for example), and taxis. I've had experiences with seeing animals about to cross the road and I was able to brake in time to avoid hitting the animal (dogs, cats, cows, horses). I was able to brake "before" they entered the road. Would a self driving car sense that these animals are about to cross the road and begin braking before they enter the road? What about road work situations? A road crew employee is in the road with a stop/slow sign. You can't safely go until he/she turns the sign to slow. How would a self driving vehicle handle such a situation? Now imagine a vehicle owner who only lets the vehicle self drive now trying to drive the vehicle without the self driving controls for the first time in a year or more. Their inexperience could result in an accident when their driving is needed.
The main danger of self-driving cars is not that cars won't be able to operate safely, but that they might be hacked. Imagine a cybercriminal pushing a GPS map update that sends thousands of cars down a cliff. Or a prankster that broadcasts an accident on a highway when there is none, just to divert traffic to other roads for his amusement.
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.
Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.
Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
).
