6 ms·

You'd think that governments would have some stake in providing infrastructure specifically for this. Underground magnets and radios along important roadways...

You'd think that governments would have some stake in providing infrastructure specifically for this. Underground magnets and radios along important roadways...that kind of thing. What I'm getting at is, there's no reason to assume that the current amount of information presented on a roadway (stripes, lights, stopsigns, etc.) is sufficient for autonomous driving. Or maybe it is, but adding a few more inexpensive things would make it that much easier to achieve.

MetaWhirledPeas 3y ago
It will need to be visual, so that people can immediately recognize compliance, in addition to machines.
- lb1lf 3y ago
  Not necessarily; I find it doubtful a person can stay alert and ready to take the wheel at a moment's notice after the car has minded its own business for however long. Hence, either /you/ drive, or /the car/ drives. If the car doesn't understand what is going on, it should stop. (Humans really suck at monitoring boring, monotonous processes. We cannot expect people to do so in a self-driving car for hour after hour on the off chance something happens.)
  - MetaWhirledPeas 3y ago
    > I find it doubtful a person can stay alert and ready to take the wheel at a moment's notice after the car has minded its own business for however long I don't mean so people can navigate; I mean so people can complain to a city or district when a road is improperly marked.
spookie 3y ago
I'm not sure if this would help. It might be somewhat inexpensive, but at scale? I don't know. Most roads aren't even properly painted ._.
- beefield 3y ago
  I'd be happy to have autonomous driving in just a few roads, namely highways. I really do not understand why there is practically zero interest in getting me out of the wheel where it would be both the easiest and the most useful. I'm more than happy to drive the bits within the city of I just could rest in between. Both embedding some kind of track in the highway my car can follow or a wireless peloton technology behind trucks should be orders of magnitude easier problem to solve than full self driving.
- ClumsyPilot 3y ago
  When I did my driving test, I have to drive through a road that was not painted. I was still expected to drive without crashing into anything.
ethanbond 3y ago
Even "inexpensive" things spread over millions of miles of roadway gets pretty expensive. Let's start with the basics of reliable painting and safe road surface.
whowe444 3y ago
One reason to assume that the current amount of information presented on a roadway is sufficient is that humans can learn to drive in a matter of hours with only two vision sensors mounted on their heads.
- mattnewton 3y ago
  The US DoT estimated $340 billion lost in traffic accident damage and just over 42,000 people’s lives lost in 2021. I think we accept that largely because it is diffused across lots of people. Self driving cars would centralize some of this liability and so they have to be an order of magnitude safer before they are accepted to replace our current system. And the existing “system” of human brains has millions of years of R&D; I think anyone in the industry should be accepting any unfair advantages we can get in the name of safety. After all does anyone care if the plane they are in is flapping its wings?
- louniks 3y ago
  Except we don't, not really. We're still leveraging 15 or so years of experience living the real world and walking on sidewalks and crossing streets and observing how other humans drive cars and being passengers in cars before we ever try driving ourselves. Not to mention, some humans definitely don't learn to drive in a matter of hours, they need weeks if not months.
  - whowe444 3y ago
    [dead]
- jvm___ 3y ago
  Plus the knowledge and intuition that, for example Firetruck >_____________< Fire Hydrant Something that looks flat but is actually a fire hose means you should stop instead of driving over it - this was actually issue in Phoenix(?) A homeless lady was killed by a self-driving car. She was walking a loaded bicycle, at 2am, on a road that doesn't usually have pedestrians. The self-driving car couldn't decide if she was a bike or a human so it didn't brake and the human-driver backup had zoned out as the car had made 99.99% of the correct decisions. A self-driving car is like a jigsaw puzzle - there's one way to do it right, and infinite ways for it to be completely or subtly wrong. "What do you see" and "What does it mean" Humans and cameras have the same "What do you see" input, but the "What does it mean" category hasn't been conquered by cars.
foobiekr 3y ago
The problem space in self driving is the set of unexpected and uncommon events. Pretty much any solution can do lanekeeping fine.
jjoonathan 3y ago
When the road quality, paint, and signage is decent, modern self driving stacks really don't have any problem seeing it and using it. Effort should be focused on improving those so that everyone benefits. If Tesla (or any self driving player, but the Tesla FSD stack drives a million miles/day across the US right now) published a road quality score, home buyers would quickly start paying attention and then home owners would quickly start paying attention. This would push forward the political will to fix the roads, at least in places where low road quality is driven by budgeting problems rather than outright poverty.
- bonestamp2 3y ago
  > When the road quality, paint, and signage is decent True, but the problem is that they're often not. That said, I'd rather see infrastructure spending on maintaining the roads, paint, and signage, than adding features specifically for autonomous driving (that will also likely be unmaintained and become unreliable over time).
  - hanniabu 3y ago
    Better to not design cars that rely on parts of the infrastructure that won't be maintained
    - bonestamp2 3y ago
      Exactly. We should only maintain the things that the people rely on, and build the cars to rely on those same things... at least for now. Once the cars are good enough that people are no longer in control of cars, then we can look at changing that standard to reduce the cost of infrastructure maintenance.
- amanaplanacanal 3y ago
  Almost nobody wants higher taxes to fix roads. Technical problems are easy compared to political problems. I think we are stuck with what we have.
hn_throwaway_99 3y ago
This would never be feasible. To start, forget underground magnets and radios. If you just started with properly painted lines, always-functioning traffic signals, and correct street signs, autonomous driving would be much, much easier than it is now. But the vast majority of the difficulty in autonomous driving is simply dealing with the messiness of the real world. For example, many of the most serious and notable autonomous driving crashes have been the result of faulty road conditions (e.g. https://www.theverge.com/2020/2/25/21153320/tesla-autopilot-walter-huang-death-ntsb-probable-cause https://www.theverge.com/2020/2/25/21153320/tesla-autopilot-... - in that case, the driver wasn't paying attention, but the fact that lane markers were worn down is what caused Autopilot to lose lane tracking, and the fact that crash attenuator in front of the barrier was damaged and not replaced contributed to the driver's death). Government can't even keep decent lane lines on the roads, and thus I think an autonomous driving system that had to depend on magnets or radios (that failed if those were out) would be even more of a non-starter.
- hermitdev 3y ago
  > If you just started with properly painted lines, always-functioning traffic signals, and correct street signs, autonomous driving would be much, much easier than it is now. If you had this, driving period would be safer, automated or not. And that's before getting into inclement weather conditions where, say, lines or signage may not be visible.

MetaWhirledPeas 3y ago

It will need to be visual, so that people can immediately recognize compliance, in addition to machines.

lb1lf 3y ago
Not necessarily; I find it doubtful a person can stay alert and ready to take the wheel at a moment's notice after the car has minded its own business for however long. Hence, either /you/ drive, or /the car/ drives. If the car doesn't understand what is going on, it should stop. (Humans really suck at monitoring boring, monotonous processes. We cannot expect people to do so in a self-driving car for hour after hour on the off chance something happens.)
- MetaWhirledPeas 3y ago
  > I find it doubtful a person can stay alert and ready to take the wheel at a moment's notice after the car has minded its own business for however long I don't mean so people can navigate; I mean so people can complain to a city or district when a road is improperly marked.

spookie 3y ago

I'm not sure if this would help. It might be somewhat inexpensive, but at scale? I don't know. Most roads aren't even properly painted ._.

beefield 3y ago
I'd be happy to have autonomous driving in just a few roads, namely highways. I really do not understand why there is practically zero interest in getting me out of the wheel where it would be both the easiest and the most useful. I'm more than happy to drive the bits within the city of I just could rest in between. Both embedding some kind of track in the highway my car can follow or a wireless peloton technology behind trucks should be orders of magnitude easier problem to solve than full self driving.
ClumsyPilot 3y ago
When I did my driving test, I have to drive through a road that was not painted. I was still expected to drive without crashing into anything.

ethanbond 3y ago

Even "inexpensive" things spread over millions of miles of roadway gets pretty expensive. Let's start with the basics of reliable painting and safe road surface.

whowe444 3y ago

One reason to assume that the current amount of information presented on a roadway is sufficient is that humans can learn to drive in a matter of hours with only two vision sensors mounted on their heads.

mattnewton 3y ago
The US DoT estimated $340 billion lost in traffic accident damage and just over 42,000 people’s lives lost in 2021. I think we accept that largely because it is diffused across lots of people. Self driving cars would centralize some of this liability and so they have to be an order of magnitude safer before they are accepted to replace our current system. And the existing “system” of human brains has millions of years of R&D; I think anyone in the industry should be accepting any unfair advantages we can get in the name of safety. After all does anyone care if the plane they are in is flapping its wings?
louniks 3y ago
Except we don't, not really. We're still leveraging 15 or so years of experience living the real world and walking on sidewalks and crossing streets and observing how other humans drive cars and being passengers in cars before we ever try driving ourselves. Not to mention, some humans definitely don't learn to drive in a matter of hours, they need weeks if not months.
- whowe444 3y ago
  [dead]
jvm___ 3y ago
Plus the knowledge and intuition that, for example Firetruck >_____________< Fire Hydrant Something that looks flat but is actually a fire hose means you should stop instead of driving over it - this was actually issue in Phoenix(?) A homeless lady was killed by a self-driving car. She was walking a loaded bicycle, at 2am, on a road that doesn't usually have pedestrians. The self-driving car couldn't decide if she was a bike or a human so it didn't brake and the human-driver backup had zoned out as the car had made 99.99% of the correct decisions. A self-driving car is like a jigsaw puzzle - there's one way to do it right, and infinite ways for it to be completely or subtly wrong. "What do you see" and "What does it mean" Humans and cameras have the same "What do you see" input, but the "What does it mean" category hasn't been conquered by cars.

foobiekr 3y ago

The problem space in self driving is the set of unexpected and uncommon events. Pretty much any solution can do lanekeeping fine.

jjoonathan 3y ago

When the road quality, paint, and signage is decent, modern self driving stacks really don't have any problem seeing it and using it. Effort should be focused on improving those so that everyone benefits. If Tesla (or any self driving player, but the Tesla FSD stack drives a million miles/day across the US right now) published a road quality score, home buyers would quickly start paying attention and then home owners would quickly start paying attention. This would push forward the political will to fix the roads, at least in places where low road quality is driven by budgeting problems rather than outright poverty.

bonestamp2 3y ago
> When the road quality, paint, and signage is decent True, but the problem is that they're often not. That said, I'd rather see infrastructure spending on maintaining the roads, paint, and signage, than adding features specifically for autonomous driving (that will also likely be unmaintained and become unreliable over time).
- hanniabu 3y ago
  Better to not design cars that rely on parts of the infrastructure that won't be maintained
  - bonestamp2 3y ago
    Exactly. We should only maintain the things that the people rely on, and build the cars to rely on those same things... at least for now. Once the cars are good enough that people are no longer in control of cars, then we can look at changing that standard to reduce the cost of infrastructure maintenance.
amanaplanacanal 3y ago
Almost nobody wants higher taxes to fix roads. Technical problems are easy compared to political problems. I think we are stuck with what we have.

hn_throwaway_99 3y ago

This would never be feasible. To start, forget underground magnets and radios. If you just started with properly painted lines, always-functioning traffic signals, and correct street signs, autonomous driving would be much, much easier than it is now. But the vast majority of the difficulty in autonomous driving is simply dealing with the messiness of the real world. For example, many of the most serious and notable autonomous driving crashes have been the result of faulty road conditions (e.g. https://www.theverge.com/2020/2/25/21153320/tesla-autopilot-walter-huang-death-ntsb-probable-cause https://www.theverge.com/2020/2/25/21153320/tesla-autopilot-... - in that case, the driver wasn't paying attention, but the fact that lane markers were worn down is what caused Autopilot to lose lane tracking, and the fact that crash attenuator in front of the barrier was damaged and not replaced contributed to the driver's death). Government can't even keep decent lane lines on the roads, and thus I think an autonomous driving system that had to depend on magnets or radios (that failed if those were out) would be even more of a non-starter.

hermitdev 3y ago
> If you just started with properly painted lines, always-functioning traffic signals, and correct street signs, autonomous driving would be much, much easier than it is now. If you had this, driving period would be safer, automated or not. And that's before getting into inclement weather conditions where, say, lines or signage may not be visible.