I think road signs will have embedded codes for self driving cars. Whether it’s a local broadcast signal, a QR code, or just extra blocky letters for the computer to read easier, road signage and signals will be directed to the computer in the car, not to inform the human in the car.
Maybe much farther out in the future when autonomous vehicles are the default. That’s a lot of signage to rejigger for very little gain, while mapping and CV already handle that small part of driving quite well.
I imagine the FSD friendly signage will eventually get slow-rolled. That is, maybe you’ll see proactive replacement on major roads and highways, but local roads won’t be updated until the sign needs to be be replaced anyway.
Maybe, but cars are already pretty good about reading signs.
It’s the lack of signs that’s the problem. Self-driving cars are pretty good on a well marked and signed road: if that’s all it took, we’re there. It’s the ubiquity of exceptions and edge cases that’s the problem.
My car recently did a one month trial of self-driving and it was a lot of fun. Also eye-opening. It did work really well on well-marked roads. However it also made me notice just how poorly marked most are. For example, it was great about staying centered between lane lines. However most local roads don’t paint the edge lines, or even the center line is worn off on many roads. Then the car is confused. I can’t even imagine what the car would do if everything were covered with snow, which does happen a lot around here
Yea. This is my big thing. Probably good on well traveled and maintained highways. Mostly good on city streets. It’s gonna be dog shit on a rural road.
I’m also curious how they would handle snow or other inclement weather that obstructs lane lines completely.
One of the most amazing things about this would be to remove signs altogether. Just embed the sensor in the pavement and give the space the signs took up back to people, nature, or literally anything else.
Huge overhead highway gantries and traffic lights would be wonderful to remove, too. City sidewalks are narrow enough as is and they would be way better without 20+ft tall metal poles jutting out of the ground. Hopefully we can put trees in their place, but maybe I’m dreaming.
Having sensors in the road (along the side and middle line) and receivers on cars would do a lot of good.
The sensors could tell the car how close it is to the side/middle. They could tell the car how close other cars are around them. They could tell the car the condition of the road surface (ex wet or icy). They could be programmed to the speed limit of the road and keep the cars moving at that speed. Probably many others uses.
The downside is building a sensor that can withstand weather, traffic, crumbling infrastructure. Also they would have to have a way to stay powered up for years. And if you made them re-programmable there is another level of security that needs to be added to them. And then there is the cost to retro-fit all the roads with these sensors and building a common receiver that could be installed on new cars.
I don’t think my idea is practical at all but it’s a cool idea
Surely there is a viable, inexpensive, simple solution. I think it would have to be a permanently programmed chip type of thing that is cheap enough that if you want to change the speed limit or lane specifications, it would be cost effective to replace all the chips along that length of road.
I think optics can make a big difference. Seeing the cameras in cars become much better so that they can read easier. But all road signage is already codified in a way that it should be relatively easy to do OCR or even matching on.
I think road signs will have embedded codes for self driving cars. Whether it’s a local broadcast signal, a QR code, or just extra blocky letters for the computer to read easier, road signage and signals will be directed to the computer in the car, not to inform the human in the car.
Maybe much farther out in the future when autonomous vehicles are the default. That’s a lot of signage to rejigger for very little gain, while mapping and CV already handle that small part of driving quite well.
Hmm. Maybe the kind of signage I’m thinking of would only be useful for construction zones and emergency vehicles.
I imagine the FSD friendly signage will eventually get slow-rolled. That is, maybe you’ll see proactive replacement on major roads and highways, but local roads won’t be updated until the sign needs to be be replaced anyway.
Maybe, but cars are already pretty good about reading signs.
It’s the lack of signs that’s the problem. Self-driving cars are pretty good on a well marked and signed road: if that’s all it took, we’re there. It’s the ubiquity of exceptions and edge cases that’s the problem.
My car recently did a one month trial of self-driving and it was a lot of fun. Also eye-opening. It did work really well on well-marked roads. However it also made me notice just how poorly marked most are. For example, it was great about staying centered between lane lines. However most local roads don’t paint the edge lines, or even the center line is worn off on many roads. Then the car is confused. I can’t even imagine what the car would do if everything were covered with snow, which does happen a lot around here
Yea. This is my big thing. Probably good on well traveled and maintained highways. Mostly good on city streets. It’s gonna be dog shit on a rural road.
I’m also curious how they would handle snow or other inclement weather that obstructs lane lines completely.
One of the most amazing things about this would be to remove signs altogether. Just embed the sensor in the pavement and give the space the signs took up back to people, nature, or literally anything else.
Huge overhead highway gantries and traffic lights would be wonderful to remove, too. City sidewalks are narrow enough as is and they would be way better without 20+ft tall metal poles jutting out of the ground. Hopefully we can put trees in their place, but maybe I’m dreaming.
I’ve thought about this idea for a long time.
Having sensors in the road (along the side and middle line) and receivers on cars would do a lot of good.
The sensors could tell the car how close it is to the side/middle. They could tell the car how close other cars are around them. They could tell the car the condition of the road surface (ex wet or icy). They could be programmed to the speed limit of the road and keep the cars moving at that speed. Probably many others uses.
The downside is building a sensor that can withstand weather, traffic, crumbling infrastructure. Also they would have to have a way to stay powered up for years. And if you made them re-programmable there is another level of security that needs to be added to them. And then there is the cost to retro-fit all the roads with these sensors and building a common receiver that could be installed on new cars.
I don’t think my idea is practical at all but it’s a cool idea
Surely there is a viable, inexpensive, simple solution. I think it would have to be a permanently programmed chip type of thing that is cheap enough that if you want to change the speed limit or lane specifications, it would be cost effective to replace all the chips along that length of road.
Ooo like higher powered rfid tags! The info could even then be relayed to the driver via the on screen display since theyre now all required.
It could be relayed inside. But in a self driving car, the driver’s seat is just a second front passenger seat
Oh wow, that does sound amazing.
I think optics can make a big difference. Seeing the cameras in cars become much better so that they can read easier. But all road signage is already codified in a way that it should be relatively easy to do OCR or even matching on.