4 ms·

Is there any talk of installing some kind of guideline along existing roads that would help navigation or vision? Something like a bright line or a buried magne

by ppod 11y ago

Is there any talk of installing some kind of guideline along existing roads that would help navigation or vision? Something like a bright line or a buried magnetic (or radio?) 'rail' for the cars to follow? I've read that vision is still a big challenge in poor conditions.

agentx3r 11y ago
At the point when the system is unable to analyze the static environment (e.g. lane markers) due to poor visibility, it's probably unable to analyze and deal with dynamic situations safely.
dclowd9901 11y ago
It's a big challenge for non-self driving cars too. If road maintenance programs aren't willing to repaint on occasion or provide lane keeping alternatives for more fallible humans, what hope do we have of them doing it for machines.
- vonmoltke 11y ago
  Indeed. Try driving in certain parts of Dallas at night or in the rain.
elchief 11y ago
Same old complaint every thread. Ford self-driving cars can drive in the snow: http://www.wired.com/2016/01/the-clever-way-fords-self-driving-cars-navigate-in-snow/ http://www.wired.com/2016/01/the-clever-way-fords-self-drivi...
- garrettgrimsley 11y ago
  It isn't fair to dismissively refer to their concern as the "same old complaint" and then link to a new story that is four days old for backing. Further, your link doesn't even sufficiently address their concern. In fact, it specifically mentions that poor weather is a challenge. >This doesn’t mean all the problems with autonomous driving in bad weather are solved. Falling rain and snow can interfere with LIDAR and cameras, and safely driving requires more than knowing where you are on a map—you also need to be able to see those temporary obstacles.
csours 11y ago
There is an industry standard suite of Vehicle communications protocols called V2X [1]. These can be used for SDC automation but no one is planning on relying on them, because there will always be places that do not have them. They can add value in some situations thought. 1. https://www.google.com/search?q=v2x https://www.google.com/search?q=v2x
Animats 11y ago
The modern version is that Volvo has proposed putting magnets, in the form of nails, into pavements. This is to provide a hint of where the lanes are under snow. It also provides guidance for snowplows. These would probably be driven in by the same machines that stripe lanes. This is an old idea from the early days of automatic driving, but today they'd just be viewed as a hint, not something to be followed blindly. There are areas where high visibility snow stakes are placed alongside highways for driver and snowplow guidance. Hokkaido puts large downward pointing arrows over snowy highways to mark the shoulder.[1] Wire guidance systems have been around for years, for industrial robot vehicles. GM put them on their test track for Firebird III in the 1960s. But it's no longer necessary to have even that much infrastructure. [1] https://www.google.com/maps/@43.9303297,143.8076743,3a,75y,340.31h,89.77t/data=!3m6!1e1!3m4!1sUXNmu36LICOLfwA2xlsBdg!2e0!7i13312!8i6656 https://www.google.com/maps/@43.9303297,143.8076743,3a,75y,3...
WalterBright 11y ago
It may be a lot cheaper and more reliable to improve the accuracy of GPS (it's a few meters at the moment.)

by ppod 11y ago

agentx3r 11y ago

At the point when the system is unable to analyze the static environment (e.g. lane markers) due to poor visibility, it's probably unable to analyze and deal with dynamic situations safely.

dclowd9901 11y ago

It's a big challenge for non-self driving cars too. If road maintenance programs aren't willing to repaint on occasion or provide lane keeping alternatives for more fallible humans, what hope do we have of them doing it for machines.

vonmoltke 11y ago
Indeed. Try driving in certain parts of Dallas at night or in the rain.

elchief 11y ago

Same old complaint every thread. Ford self-driving cars can drive in the snow: http://www.wired.com/2016/01/the-clever-way-fords-self-driving-cars-navigate-in-snow/ http://www.wired.com/2016/01/the-clever-way-fords-self-drivi...

garrettgrimsley 11y ago
It isn't fair to dismissively refer to their concern as the "same old complaint" and then link to a new story that is four days old for backing. Further, your link doesn't even sufficiently address their concern. In fact, it specifically mentions that poor weather is a challenge. >This doesn’t mean all the problems with autonomous driving in bad weather are solved. Falling rain and snow can interfere with LIDAR and cameras, and safely driving requires more than knowing where you are on a map—you also need to be able to see those temporary obstacles.

csours 11y ago

There is an industry standard suite of Vehicle communications protocols called V2X [1]. These can be used for SDC automation but no one is planning on relying on them, because there will always be places that do not have them. They can add value in some situations thought. 1. https://www.google.com/search?q=v2x https://www.google.com/search?q=v2x

Animats 11y ago

The modern version is that Volvo has proposed putting magnets, in the form of nails, into pavements. This is to provide a hint of where the lanes are under snow. It also provides guidance for snowplows. These would probably be driven in by the same machines that stripe lanes. This is an old idea from the early days of automatic driving, but today they'd just be viewed as a hint, not something to be followed blindly. There are areas where high visibility snow stakes are placed alongside highways for driver and snowplow guidance. Hokkaido puts large downward pointing arrows over snowy highways to mark the shoulder.[1] Wire guidance systems have been around for years, for industrial robot vehicles. GM put them on their test track for Firebird III in the 1960s. But it's no longer necessary to have even that much infrastructure. [1] https://www.google.com/maps/@43.9303297,143.8076743,3a,75y,340.31h,89.77t/data=!3m6!1e1!3m4!1sUXNmu36LICOLfwA2xlsBdg!2e0!7i13312!8i6656 https://www.google.com/maps/@43.9303297,143.8076743,3a,75y,3...

WalterBright 11y ago

It may be a lot cheaper and more reliable to improve the accuracy of GPS (it's a few meters at the moment.)