Self-Adapting Traffic Lights

Roland Piquepaille writes "If you're like me, I bet you hate moments when you're in a hurry and all the traffic lights seem to intentionally switch to red just in front of your car. Now, according to Nature, a Belgian traffic researcher thinks that traffic lights that respond to local conditions could ease congestion and reduce your frustration. His method would not give you the individual power to switch the light to green. But if you were part of a group of cars approaching a red light, inexpensive traffic-flow sensors would detect your group in advance and turn the light to green. His simulations show that such adaptive traffic control is 30% more efficient than traditional ways of regulating traffic. However, his system has not been adopted by any large city. So you'll continue to be frustrated by these ?%&$! traffic lights for a while. You'll find more details and references in this overview."

Comments

Weight Sensors

[fembots]

My city has weight sensors laid under the tarmacs, so it knows if there are cars waiting/approaching and switches lights accordingly, or if it shall let the other direction keep going.

The real problem only arises when there are too many cards coming from all directions, and the lights will switch to the "traditional method" that is based on a predefined interval.

It's a catch-22 - Gershenson admits that the benefits wouldn't be as large in a big city where the situation is much more complex than in his simulations, however only bigger city needs to/will consider such traffic control.

Re:Weight Sensors

[Anonymous Coward]

Are you sure they are weight sensors? I was under the belief that those sensors worked on the principle of induction. They send a magnetic field up through the pavement, which induces a current in any metal vehicle above. That induced current, in turn, creates a magnetic field which is sent back down through the pavement to the sensor. Works in any temperature and will even dedict lightweight objects like motorcycles.

Re:Weight Sensors

[rawket.scientist]

will even dedict lightweight objects like motorcycles

But not, alas, bicycles. There's one redlight back at my alma mater that doesn't turn unless you trip the sensor; it was either run it, or wait half an hour for a car to show up.

Re:Weight Sensors

[Mundocani]

Try lining up your bike tires with one of the edges of the sensor when you pull up. Once I learned to do this I was able to reliably trip the sensor and get the light to change.

The sensors work off of magnetic induction (like a metal detector) and your bike just doesn't have that much metal to be detected. Positioning yourself along the edge of the road's sensor should trigger the light.

Re:Weight Sensors

[locnar42]

A trick I learned a long time ago is to shutoff and restart the motorcycle. The starter motor has enough electrical force to trigger most sensors.

My current bike won't engage the solenoid if the bike is already running. I don't have to shutoff the bike first. I just press the start button real quick and the light turns green. Since most lights are switching to camera sensors this isn't as useful now. Flash the lights a couple of times and the camera will pick it up as motion.

Re:Weight Sensors

[chameleon3]

Well, they have the capability to detect motorcycles and bicycles, but often the street dept. lowers the sensitivity to the point that only vehicles larger than a small car are detected. This is well documented on many motorcycle discussion boards.

Motorcyclists in Tennessee can legally run red lights because of this

http://www.motorcyclecruiser.com/newsandupdates/TN redlitelaw/

Re:Weight Sensors

[sylvester]

Try lining up your bike tires with one of the edges of the sensor when you pull up. Once I learned to do this I was able to reliably trip the sensor and get the light to change.

In Ottawa, Canada, the majority of lights have sensors, and the vast majority of those sensors are marked with three dots that indicate where bicycles should go to have the best chance of triggering the sensors. The sensors that have those dots are very reliably tripped.

-Rob

Re:Weight Sensors

[Moraelin]

Yet it never seemed to occur to you to just obey the speed limit...

The thing about timed streetlights is that they're calibrated for a given speed. If they're, say, calibrated for 30mph and they're 1/4 of a mile apart, they'll turn green every 30 seconds regardless of your speed. If you go "only" 5-10 MPH faster... you just catch a red light, and still don't get home any faster than someone who obeys the speed limit.

I.e., you'd think people would get the idea already that there is really no reward for endangering everyone around. Someone who stuck to the speed limit got home in exactly the same time, and obviously with less stress. Didn't need to use up extra gas accelerating and decelerating all the time either.

And yes, I do mean endangering. Due to the elementary physics fact that kinetic energy is proportional to the square of the speed, so is the braking distance. E.g., the speed difference between 50 km/h and 70 km/h is 40%, but the braking distance _doubles_.

Add poor visibility at night (you might not see a kid dashing to cross the street until he's in front of your beams), the driver _and_ everyone around being tired, etc, and I really _don't_ need people doing "only" 10mph over the limit at night.

And again, as you've noticed, it doesn't even get you home faster. It just makes you stop at the next red light.

But naah... for some people speeding is like _the_ proof of their manhood. Obeying the traffic laws or not driving like an irresponsible maniac, that's like admitting sexual impotence. Or worse.

Geesh.

Traffic Lights

[AetherGoth]

When I was in San Jose, CA a couple of years back, they had a system up where sensors in the road would pick up cars at intersections. They then used microwave antennae to broadcast the information to lights further down the road. So if you were driving along at night with nobody else on the road, you would get long strings of green lights going your way.

Re:Traffic Lights

[COskigrl]

"a Belgian traffic researcher thinks that traffic lights that respond to local conditions could ease congestion and reduce your frustration. His method would not give you the individual power to switch the light to green. But if you were part of a group of cars approaching a red light, inexpensive traffic-flow sensors would detect your group in advance and turn the light to green. However, his system has not been adopted by any large city."

I am a traffic engineer, and traffic lights similar to what this Belgian traffic researcher describes already exist. They are called "actuated signals." They work as follows: Loops (not weight sensors, but magnetic loops) are placed in the roadway approx. 300 meters before the traffic light, then 200m, 100m, 50m and 10m. When the light is green for this path, every time a car drives over a loop(assume 300m loop), the green light time is extended long enough for the car to reach the next loop (200m), and so on and so forth until it reaches the 10m loop, where the green light is extended long enough for the car to travel safely through the intersection. Now, if the 300m loop is not reactivated every 3 seconds, the light "times-out" and will turn red once all vehicles have passed through the intersection safely (so if a vehicle is on the 100m loop, the light doesn't just turn red). Additionally, the light has a maximum green cycle time (sum of green and yellow light time), typically 58 seconds. So, if there is a never-ending stream of cars, the light doesn't remain green forever. I hope this clears things up a bit. Also, actuated signals are intended for minor arterials (major collector streets), not for principal arterials (expressways and large intersections).

"His method would not give you the individual power to switch the light to green."

Actuated signals do give you the individual power to switch the light to green.

Traffic Simulations

[lonedfx]

His simulations show that such adaptive traffic control is 30% more efficient than traditional ways of regulating traffic. However, his system has not been adopted by any large city

I'll guess that the reason why is because a simulation shows this, not a real test. Traffic simulation has been a topic of much research, but as far as I'm aware, little convincing results have emerged... Simulations based on liquid flow do not work (they do not give anything like an average traffic), and those based on drivers modelization (ie, x % of 'aggressive drivers', y % of 'sloppy drivers', z % of 'careful drivers' etc) become incresingly complex and demanding with the scale of the simulation... I'm not aware of anything practical ever done with these (feel free to correct me).

In any case, if his adaptive system does work, it's a breakthrough. I've worked a few years back with people in charge of traffic and roads around Paris, and from what I've been told, nothing like this has ever worked better than static programming (with the exception of multiple programmings for different time of the day). From what I remember, even getting such programming right demands extremely experienced people. Of course, this might be specific to Europe where intersections are rarely perpendicular and often involve "creative" solutions.

SCATS

[Anonymous Coward]

I'm from a city in New Zealand (Hamilton) where we have a self monitoring system tied back to the City Council. Unfortunately it seems to be a little too smart, holding patterns that don't reflect the traffic. As a result, traffic changes its flow each day (drivers choose new routes) which further changes the trending and thus cancels the advantages you'd hope to gain. When the system is out or loops are cut (roadworks) the system reverts to timers/loops which seem to work better. Perhaps it's just when we add users the perfect system suddenly becomes imperfect...

Traffic Calming

[Nonesuch]

Most of the "traffic calming" enhancements to signals and lane markings are meant to slow down traffic, or even to introduce delays intended to drive commuters out of individual cars and into mass transit (See AATC).

It's nice to see a traffic signal enhancement that will actually make driving more efficient and direct rather than the opposite.

Motorcycles

[wpc4]

Now I just need one that will recognize my motorcycle at 2am when no cars are around to tigger the lights for me.

I ride a supersport Yamaha YZF-R6. Weighs about 410 wet and I have problems triggering many stop lights, so much so that I have areas I don't ride when traffic is light because they never turn green for me.

Re:Motorcycles

[morcheeba]

Try this - a magnet designed to induce voltage across the coils as you move across them. Depending on how the sensor is designed, this voltage may trigger it (otherwise, it would be using the inductance change from the frame of a car)

Not a chance

[LanceUppercut]

US is currently preparing for a completely different thing - a more or less massive roll-out of red-light cameras (the thing where you get you car's photograph in the mail and a red-light ticket). As a preparation for this measure, stop lights are adjusted (most of the time the duration of yellow is simply reduced) in order to increase you chances of running red light, thus increasing the profit generated by red-light tickets. (This will also increase the fatality rate, of course, but this doesn't seem to be a reason for concern.) I'm amazed how much more often these days in California I see cars crossing intersection right under my nose even when I have green. A couple of years ago I'd see something like this about once in a month. These days I see it virtually every day. In this evironment it is highly unlikely (read - impossible) that US authorities will implement anything tha will to decrease your chances of running red light. Today they prefer to make money by decresing public safety, not by increasing it. So you can forget about anything like "driver friendly" stop light for a while.

Here is why these aren't on your street

[Rheagar]

Assume (for this post) that these devices actually work and improve the efficiency of city streets. This is why you can't find them on your block:

(1) Safety. A lot of effort is spent proving that a traffic control device is safe. When traffic lights screw up and allow opposing green lights, people die. It is entirely unacceptable for a traffic control device to screw up.
So when a city is faced with buying a proven design or a new advanced design that improves efficiency but may be a liability concern, the city will go with the proven design.
I concede that the new system would be tested endlessly, but I claim that any complex system will have flaws that don't show up until deployed in the field. I've seen unbreakable unix systems crash. It happens.
I think that provable safety in this application can (and will someday) be done. I just wouldn't want to be the first city adopting it.
So another option to ensure safety is redundancy such as that used in some airplanes. That is, multiple independant systems working on the traffic problem, and if any of them fail the others will notice. Doing this right costs money, which brings us to point 2.

(2) Cost. My city really doesn't even bother fixing road problems. I went to Berkeley CA the other day and they had enormous potholes that were "fixed" by painting bright colors around them so they could be avoided. If Berkeley doesn't want to spend a couple bucks to patch a hole, then why would your little town bother to consider removing existing systems that cost hundreds of thousands of dollars and replace them with ones that probably cost more?
Further, why would you want your city to spend this money for a marginal improvement in flow? The answer is because some intersections are so terrible that you always are caught up in traffic. These intersections are the bottlenecks that hold up everybody, ones where 30% improved efficiency would be a blessing, which brings us to the 3rd point.

(3) This doesn't help the worst intersections. This switching system would be nice for those pesky lights in your neighborhood that always seem to be red when you arrive, and that you are always first in line and usually the only one to go through in your direction.
I claim that the intersections which could use a 30% improvement the most are those that would not be helped by this system. That is because no matter which side is getting green, every precious second of green light is being used by traffic. This is 100% efficiency, as measured by throughput / theoretical maximum throughput. You can not improve this system by watching for groups of cars, since there are always groups of cars coming.
This would be a neat feature on some intersections, but these intersections aren't the ones that DOT really focuses on improving. The effort involved in making small intersections intelligently switch lights isn't generally worth the cost of doing so.

That said, I'd like to see this in use in my neighborhood, and I'm glad that people are looking into solving traffic congestion problems.

Sydney has had this for the last 20 years

[goodie]

Sydney has had this for the over 20 years.

Around 2500 of the intersections in Sydney are linked together and they "marry" and "divorce" each other based on live statistical data as cars flow through the intersections.

It's a self-calibrating system. It has been exported to many countries.

The local intersection controllers measure traffic flows and adjust timings locally and also are linked to regional controllers that share statistics for an area and these regional controllers are all linked to the central monitoring facility in the city.

Google on the Sydney Co-ordinated Adaptive Traffic System (SCATS)

e.g. http://www.traffic-tech.com/pdf/scatsbrochure.pdf