Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Study: Traffic lights should respond to cars, not the other way around

sciencenews.org

Post reply on HN
Print view View original

51–60 of 75 posts

Re: Study: Traffic lights should respond to cars, not the other way around

#51

Post by kd0amg »

njharman wrote: post #37
krschultz wrote: post #11

Not to sound like a radical libertarian but imagine a world in which roads were operated with some sense of incentives to actually MOVE TRAFFIC THROUGH. Lets have Tolls - but you get a refund when you are delayed. I imagine construction would suddenly be done 24/7 and be finished in half the time. Traffic light systems would be heavily automated/AI so that there isn't much traffic. Instead they give out construction…

Almost every place I've lived in the US construction does occur 24 hours, but maybe not 7. Commute and daily traffic require it. Maybe you're just not out 24/7 to see it? Most traffic systems are automated and their goal is to MOVE TRAFFIC THROUGH. The minimal use sensors in the roads to help determine when to switch lights. The most advanced are integrated and use many metrics and algorithms in controlling traffic s…

Almost every place I've lived in the US construction does occur 24 hours, but maybe not 7. Commute and daily traffic require it. Maybe you're just not out 24/7 to see it?

There have been a lot of places/projects where all I ever saw were closed lanes with some inactive construction equipment parked nearby. Last month was the first time I ever saw a construction project with crews working all three shifts -- it was quite the novelty, since I had never even heard of such a thing before.

Re: Study: Traffic lights should respond to cars, not the other way around

#52

Post by stretchwithme »

krschultz wrote: post #11

Not to sound like a radical libertarian but imagine a world in which roads were operated with some sense of incentives to actually MOVE TRAFFIC THROUGH. Lets have Tolls - but you get a refund when you are delayed. I imagine construction would suddenly be done 24/7 and be finished in half the time. Traffic light systems would be heavily automated/AI so that there isn't much traffic. Instead they give out construction…

Incentives are great but will be lost on bureaucrats. Its not their money. Now if it were a private system, that would be another matter

But there would be plenty of incentive to move traffic through as fast as possible, just as there is to move people through theaters and restaurants as fast as possible.

And, as with theaters and restaurants and motels and many other things, private roads would charge you more at peak times. People would try to travel at cheaper times if possible and try to reschedule activities. Traffic would actually keep flowing at the pace that gets the most cars through the network.

And this would actually reduce the amount of road building required. People would have more incentive to take mass transit. And we'd have less smog and less stress because people wouldn't be stuck in traffic on a daily basis.

And we could lower taxes. Road building is expensive and so is defending the oil. People who use the roads the most would be the ones paying for them.

Re: Study: Traffic lights should respond to cars, not the other way around

#53

Post by brc »

SeanLuke wrote: post #50

A sort of vague way to do the green wave, but okay I guess. But -- and I say this in all seriousness -- one of my PhD graduate students had a far more interesting and elegant technique five years ago. In short: each car has a piggy bank. As you're sitting at a red traffic light, it dispenses tokens into your piggy bank. When you go through a green traffic light you have to give some tokens back. Here's the catch: whe…

It's a good idea but would really require the car to have interaction with the system, or the system to read and interpret individual cars (to assign the token).

I think the advantage with the system shown is they are only measuring throughput rather than keeping tabs on individual cars. Thus I assume it's an easier problem to solve.

Re: Study: Traffic lights should respond to cars, not the other way around

#54

Post by blaines »

Where I live there are roads to nowhere and traffic lights in quiet neighborhood intersections. I'm not sure who 'engineered' the streets, but one car can stop a swarm of traffic, and the speed limits are many times ridiculously low i.e. 35mph on a two lane street in a non-residential area, high visibility, and no intersections within 2mi.

This is a broken system in which much of my time is wasted daily, and the "green wave" is not the system's goal. It's nice to see some people experimenting to find better ways to regulate traffic. It should be obvious that 90% of the time the current flow is not the calculated daily average.

After all, Since when do 50,000 cars/hour drive on a street at 2am?

Re: Study: Traffic lights should respond to cars, not the other way around

#55

Post by blaines »

eru wrote: post #49
kes wrote: post #33

Earlier quoted context omitted.

I live in a small midwestern town that has recently taken to the idea of roundabouts. At first, I was extremely pleased (having good experiences with them in Paris/French HW systems) but I quickly became discouraged when I realized that 90% of drivers had no idea what they were doing. I now avoid certain routes through town because I'm afraid of getting in an accident.

What's so hard to get for those people? Roundabouts seem simple enough to me (though I may be biased, coming from Europe and now living in Britain).

The roundabouts near me seem to be working well...

Re: Study: Traffic lights should respond to cars, not the other way around

#56

Post by points »

eru wrote: post #49
kes wrote: post #33

Earlier quoted context omitted.

I live in a small midwestern town that has recently taken to the idea of roundabouts. At first, I was extremely pleased (having good experiences with them in Paris/French HW systems) but I quickly became discouraged when I realized that 90% of drivers had no idea what they were doing. I now avoid certain routes through town because I'm afraid of getting in an accident.

What's so hard to get for those people? Roundabouts seem simple enough to me (though I may be biased, coming from Europe and now living in Britain).

Stop. Go.

They're so used to driving in straight lines. Their cars are simply not made for corners.

I guess you need to build roundabouts, update the cars so they can corner, and educate the public on how to steer.

Re: Study: Traffic lights should respond to cars, not the other way around

#57

Post by jules »

mmt wrote: post #42
brudgers wrote: post #20

Earlier quoted context omitted.

What matters is moving people, not traffic. http://commontragedies.files.wordpress.com/2009/01/cars-vs-b...

That's a bare assertion, and the link doesn't support it. There's a very important vehicle missing from the list: trucks. Add "cargo" to "people" and, suddenly, it's no longer easy to measure. How many cars are carrying cargo in addition to persons, thereby being partially a truck in function?

How many car trips include more cargo than can be carried on a bus or on a bike?

Re: Study: Traffic lights should respond to cars, not the other way around

#58

Post by khafra »

krschultz wrote: post #11

Not to sound like a radical libertarian but imagine a world in which roads were operated with some sense of incentives to actually MOVE TRAFFIC THROUGH. Lets have Tolls - but you get a refund when you are delayed. I imagine construction would suddenly be done 24/7 and be finished in half the time. Traffic light systems would be heavily automated/AI so that there isn't much traffic. Instead they give out construction…

I don't know if I would want my driving managed by Comcast.

Re: Study: Traffic lights should respond to cars, not the other way around

#59

Post by khafra »

brc wrote: post #53
SeanLuke wrote: post #50

A sort of vague way to do the green wave, but okay I guess. But -- and I say this in all seriousness -- one of my PhD graduate students had a far more interesting and elegant technique five years ago. In short: each car has a piggy bank. As you're sitting at a red traffic light, it dispenses tokens into your piggy bank. When you go through a green traffic light you have to give some tokens back. Here's the catch: whe…

It's a good idea but would really require the car to have interaction with the system, or the system to read and interpret individual cars (to assign the token). I think the advantage with the system shown is they are only measuring throughput rather than keeping tabs on individual cars. Thus I assume it's an easier problem to solve.

Yeah, for a lesser investment than that, they could put some basic sensors on so I'm not waiting with 10 other cars while a green light stays on 30 seconds in the other direction for absolutely nobody. "Improving traffic lights" is pretty low-hanging fruit.

Re: Study: Traffic lights should respond to cars, not the other way around

#60

Post by khafra »

ghjklkjh wrote: post #27
points wrote: post #15

Radical, but there's only one word needed here. ROUNDABOUT. Build some.

But four-way stops give the Americans so many opportunities to demonstrate their well known diffidence and thinking of other drivers.

s/diffidence/deference
Post reply on HN