You had to physically solder the motherboard to get lower-case on the Apple ][+. I remember my dad doing it, and it only worked in AppleWord if I recall.
Because browsers do a lot of things but they do nothing real well. Now I know a lot of apps are nothing more than a browser window hard wired to a certain website, but a well done native app will be better and smoother than a browser app every time.
But what you are suggesting is not the same thing. What you are suggesting is more like having lanes for people who own Lexus', Mercedes, and Porsches only.
I think there will be more than a few cars rolled over by front end loaders because the front end loader didn't understand what the AI was trying to do and vice-versa.
It might work at 50% *if* automated cars get a dedicated lane. This isn't always possible, and even if it is, there will be a revolt over giving preferential treatment to the people who can afford them.
Thinking about it more, one of the most difficult things to do in the winter is driving down a road that is in the process of being snow cleared. There may be large machines driving around to avoid, with windrows of ice here and there. The machines may be gone but left the windrows for another machine to pick up later. Sometimes they just forget the windrows. My wife almost hit one last winter, she was in a roundabout and they just forgot a large pile of snow in the road which she had to avoid. Autonomous vehicles will have to deal with these totally unexpected things that happen in real life. They will need to anticipate snow piles and machines and other cars and calculate some sort of a micro-rerouting that is a safe route to take. These aren't small problems, but they happen every winter.
They're testing it in A TOWN BUILT FOR SELF DRIVING VEHICLES. This means that the stop signs they are alternatively taking queues from are all there and not already bent over from a previous accident; something that is quite common to see around where I live. Also if the picture is an indicator, that is pretty tame snow. Around here, there are ice ruts, and not every street and back alley will be accurately mapped. Not sure how capable this 3D scanning is, but it will be absolutely critical that the vehicle scan the contour of the road and match the tires to the ruts. Also, no mention in this article if the cars can deal with slick driving conditions while still maintaining a decent speed, not 20 mph. Even in the article they indicate that there is a long way to go such as anticipating people which I already mentioned. All of this needs to be affordable for the average person, and vehicle manufacturers won't want to spend more than around $500 per vehicle on sensors.
To my knowledge they haven't even been tried in the snow yet, without visible lane markings. That's a fairly big challenge on its own, and probably ten times more difficult than anything they have done so far. Also current autonomy doesn't really understand the world around it; for it to be safe it will actually understand what certain objects are. For example, determine the object moving in front is a rabbit and simply avoid it and carry on, or determine the object is a ball (CHILD'S TOY) and anticipate that a child may be running after it. They are nowhere near that yet. They can make it look like they have 80 percent of the problems solved but there is 20% of edge cases that will be the hardest to resolve and the most expensive. Until they start tapping into those things, AI won't be safe.
There will only be major efficiency gains in commute times once 99% of all people have autonomous cars, and that may never be a financial reality. Manual cars will always be cheaper, so people will always have them.
The thing you forget is that humans can always choose to drive slower and more carefully. We lose that control with automation, and in that case if the automation screws up then it is responsible for a death of that person who would have otherwise been more careful.
That's the point.. there can be no source.. Unless someone figures out what human accidents are only in the exact places that Autopilot works, doung only thing things that Autopilot does. Otherwise you can't compare them. Of course a human in a blizzard pulling out of their driveway is going to get in more accidents than Autopilot in new perfect conditions that it is limited to. Even that may not be an accurate comparison because Autopilot has a human (who is now shit scared because of the previous accidents) to take over for it.
I haven't seen an extrapolation of human accidents in places that autonomous cars work for an apples to apples comparison. Right now it's hard to say because Autopilot only works in the safest situations, so comparing all human accidents to Autopilot accidents is not an accurate comparison.
Slashdot Mirror
SIGNAL RECEIVED .... Skynet connectng to drone dragnet .... SIGNAL COMPLETED
I'm going to leave this conversation now, I'm afraid my geek will show too much.
Eventually I got good at knowing where to use a single hole punch to make the notch.
You had to physically solder the motherboard to get lower-case on the Apple ][+. I remember my dad doing it, and it only worked in AppleWord if I recall.
So they had to remove the headphone jack so they could make the phone vibrate? The barometer sounded like a better idea!
That sounds a lot harder than playing Quake in your parents basement.
That doesn't work the other way around.
Because browsers do a lot of things but they do nothing real well. Now I know a lot of apps are nothing more than a browser window hard wired to a certain website, but a well done native app will be better and smoother than a browser app every time.
But what you are suggesting is not the same thing. What you are suggesting is more like having lanes for people who own Lexus', Mercedes, and Porsches only.
..and I can guarantee you there are lawyers salivating at the thought of suing automakers for it.
I think there will be more than a few cars rolled over by front end loaders because the front end loader didn't understand what the AI was trying to do and vice-versa.
It might work at 50% *if* automated cars get a dedicated lane. This isn't always possible, and even if it is, there will be a revolt over giving preferential treatment to the people who can afford them.
Thinking about it more, one of the most difficult things to do in the winter is driving down a road that is in the process of being snow cleared. There may be large machines driving around to avoid, with windrows of ice here and there. The machines may be gone but left the windrows for another machine to pick up later. Sometimes they just forget the windrows. My wife almost hit one last winter, she was in a roundabout and they just forgot a large pile of snow in the road which she had to avoid. Autonomous vehicles will have to deal with these totally unexpected things that happen in real life. They will need to anticipate snow piles and machines and other cars and calculate some sort of a micro-rerouting that is a safe route to take. These aren't small problems, but they happen every winter.
They're testing it in A TOWN BUILT FOR SELF DRIVING VEHICLES. This means that the stop signs they are alternatively taking queues from are all there and not already bent over from a previous accident; something that is quite common to see around where I live. Also if the picture is an indicator, that is pretty tame snow. Around here, there are ice ruts, and not every street and back alley will be accurately mapped. Not sure how capable this 3D scanning is, but it will be absolutely critical that the vehicle scan the contour of the road and match the tires to the ruts. Also, no mention in this article if the cars can deal with slick driving conditions while still maintaining a decent speed, not 20 mph. Even in the article they indicate that there is a long way to go such as anticipating people which I already mentioned. All of this needs to be affordable for the average person, and vehicle manufacturers won't want to spend more than around $500 per vehicle on sensors.
To my knowledge they haven't even been tried in the snow yet, without visible lane markings. That's a fairly big challenge on its own, and probably ten times more difficult than anything they have done so far. Also current autonomy doesn't really understand the world around it; for it to be safe it will actually understand what certain objects are. For example, determine the object moving in front is a rabbit and simply avoid it and carry on, or determine the object is a ball (CHILD'S TOY) and anticipate that a child may be running after it. They are nowhere near that yet. They can make it look like they have 80 percent of the problems solved but there is 20% of edge cases that will be the hardest to resolve and the most expensive. Until they start tapping into those things, AI won't be safe.
There will only be major efficiency gains in commute times once 99% of all people have autonomous cars, and that may never be a financial reality. Manual cars will always be cheaper, so people will always have them.
The thing you forget is that humans can always choose to drive slower and more carefully. We lose that control with automation, and in that case if the automation screws up then it is responsible for a death of that person who would have otherwise been more careful.
Unless you were going to watch entertainment for that 90 mins a day anyway, that's just time wasting; not gaining productivity.
That's the point.. there can be no source.. Unless someone figures out what human accidents are only in the exact places that Autopilot works, doung only thing things that Autopilot does. Otherwise you can't compare them. Of course a human in a blizzard pulling out of their driveway is going to get in more accidents than Autopilot in new perfect conditions that it is limited to. Even that may not be an accurate comparison because Autopilot has a human (who is now shit scared because of the previous accidents) to take over for it.
Absolutely, but from what I see it's around 40-50 years out.. so not sure if it is work even talking about.
Let's face it, with the state of self driving today there is plenty of reason to be anxious.
On the other hand, when it is truly autonomous, then people will use the time for something. It's just that we are very far off still.
I actually meant to reply to this... what stats are you looking at in comparison to autonomous cars?
I haven't seen an extrapolation of human accidents in places that autonomous cars work for an apples to apples comparison. Right now it's hard to say because Autopilot only works in the safest situations, so comparing all human accidents to Autopilot accidents is not an accurate comparison.
..in fifty years or so.