Article: When a signal goes on too long

[Duke87]

I encountered a stuck signal in The Bronx once. East Tremont Ave and some minor side street (I forget exactly which, it was between The Hutch and The Bruckner). The next light down the road cycled through twice, mine didn't change. So, I waited for an opening and made the left turn I had intended to make anyway through the red light. It should be noted that right on red is illegal in New York City, so that could not have been taken advantage of here.

Even more weird, though, I've encountered a couple of signals where both the red and green were lit simultaneously. One at a school driveway which was closed to vehicles because graduation ceremony was in progress. Due to heavy pedestrian and vehicle traffic, there was a cop in the intersection directing people. I assume the light was intentionally deactivated for this and through some programming oddity it ended up like that rather than dark or flashing.

The other was a case where the signal was operating normally, but one signal head facing each way had the red light seemingly stuck on. So traffic coming up or down the road would encounter two red lights when the light was red, but a red light and two green lights when the light was green. Drove passed it three times that day with it in the same condition, but didn't see if it was also happening to the side street. This was during and after a nasty storm, and it was fine when I drove past it earlier that day before the storm, so I assume lightning was involved.

[agentsteel53]

speaking of light deactivations ... I was at an event downtown last weekend with some streets closed off, and some of the traffic lights were going on their normal cycle, while others were flashing all-red. 

it was a moot point, because the streets were barricaded off, with police located strategically.

[roadman65]

Florida has one in Orlando near the Florida Mall.  Orange Blossom Trail and LaQuinta Drive when a SB vehicle on OBT turns left onto EB LaQuinta it does not give the through green movement on NB OBT.  What it does is turns green again on LaQuinta Drive.  So you have the light turn green twice during your one time red stop only if another car does not make a left again SB to EB.

Plus the signal is defaulting to timer on LaQuinta cause the detector hoops are not working.  It ends up you could wait a few minuets to travel there when this happens.  Of course, this was happening months ago and just happened again last week.  Maybe it was never fixed because no one bothered to report the signal to Orange County who operates the light.  Being it took months for people in my subdivision to make a complaint that the signal outside our development was timed wrong, I can believe that.  For almost two years people were waiting two minuets or more to exit the housing area on to our main street during the 5 AM hour, and no one thought to pick up the phone and call until earlier this year.

[deathtopumpkins]

This brings up an interesting question. Do any states have a specific time limit after which a motorist is legally allowed to assume a signal is malfunctioning and disobey it?

It's not exactly the same case, but Wisconsin makes an allowance for 2-wheeled vehicles:

Notwithstanding subd. 1., a motorcycle, moped, motor bicycle, or bicycle facing a red signal at an intersection may, after stopping as required under subd. 1. for not less than 45 seconds, proceed cautiously through the intersection before the signal turns green if no other vehicles are present at the intersection to actuate the signal and the operator of the motorcycle, moped, motor bicycle, or bicycle reasonably believes the signal is vehicle actuated. The operator of a motorcycle, moped, motor bicycle, or bicycle proceeding through a red signal under this subdivision shall yield the right-of-way to any vehicular traffic, pedestrian, bicyclist, or rider of an electric personal assistive mobility device proceeding through a green signal at the intersection or lawfully within a crosswalk or using the intersection. This subdivision does not affect any authorization for a bicyclist under subd. 2.

Other than that, I can't seem to find anything related to a malfunctioning signal in our statutes. One would hope to assume common sense would prevail, Read: Check for conflicting traffic and go when the way is clear.

This is common - I remember Virginia passed such a law a few years before I moved.

One would assume that common sense would prevail, yes, but one should know otherwise. 

[Mr_Northside]

It's not exactly the same case, but Wisconsin makes an allowance for 2-wheeled vehicles:

Notwithstanding subd. 1., a motorcycle, moped, motor bicycle, or bicycle facing a red signal at an intersection may, after stopping as required under subd. 1. for not less than 45 seconds, proceed cautiously through the intersection before the signal turns green if no other vehicles are present at the intersection to actuate the signal and the operator of the motorcycle, moped, motor bicycle, or bicycle reasonably believes the signal is vehicle actuated. The operator of a motorcycle, moped, motor bicycle, or bicycle proceeding through a red signal under this subdivision shall yield the right-of-way to any vehicular traffic, pedestrian, bicyclist, or rider of an electric personal assistive mobility device proceeding through a green signal at the intersection or lawfully within a crosswalk or using the intersection. This subdivision does not affect any authorization for a bicyclist under subd. 2.

This is common - I remember Virginia passed such a law a few years before I moved.

Good to know... after the fact.  I had a situation (in VA) last week in a left-turn lane with a left-turn signal.  I knew my bicycle would not activate it, so I just turned left when there was no oncoming traffic (and cross-traffic still had a red).  I wasn't sure if there was a law/regulation dealing with that, but wasn't going to wait for someone else to turn left behind me.

Fixed your quote - DTP

[mjb2002]

The RYG in a certain part of Aiken County is downright awful. At CHARLESTON HY's (US 78's) westernmost terminus stays red for about two minutes before it changes. That absolutely snarls up traffic on EAST PINE LOG RD & RUDY MASON PKWY (SC 118/302) at their shared intersection with WAGENER RD (SC 4/302).

No wonder why so many people who are short on time go to WILLISTON RD (US 278) and then turn right in New Ellenton on SOUTH WHISKEY RD (SC 19) to go to Aiken.

[KEK Inc.]

After a minute.  No car, no cop, no stop. 

I believe it's actually legal for motorcycles to run reds in some states if they've waited at an empty light for more than a minute.  The coil detectors don't always pick up motorcycles, and the IR detector is relatively new (and doesn't work well in weather). 

Most signals, if designed by a decent engineer, should have 20 second timer to change if it doesn't recognize any vehicles.

[Alps]

Most signals, if designed by a decent engineer, should have 20 second timer to change if it doesn't recognize any vehicles.

Wait, what? You're saying the signal should change if there's no detection? The whole idea behind full actuation is the ability to skip unused phases and "rest in green" if uncalled.

[Duke87]

I think he's talking about having loops on both roads and having the signal change every so often anyway if it detects no vehicles in either place.

Don't think that's typical, though. Usually there are no loops on the main road.

What you do want to do, however, is set it up so that if one of the loops breaks, it constantly triggers rather than going dead. Better to have a phase every cycle than to eternally skip it.

[Alps]

I think he's talking about having loops on both roads and having the signal change every so often anyway if it detects no vehicles in either place.

Don't think that's typical, though. Usually there are no loops on the main road.

What you do want to do, however, is set it up so that if one of the loops breaks, it constantly triggers rather than going dead. Better to have a phase every cycle than to eternally skip it.

Yes, and that is good traffic engineering, absolutely. But still, proper engineering for actuation when circuits are working is to rest in one phase, not cycle through. (Fixed timing doesn't get much use anymore - I'd say mainly if you're talking about a circuit of urban signals with central management.)

[Road Hog]

Few traffic signals use loops embedded in the road anymore. The ones in Texas use optic motion sensors mounted on the crossbar the last 10 years. A lot cheaper to replace when they malfunction. Doesn't reduce the rate of malfunction, unfortunately.

[Brandon]

Few traffic signals use loops embedded in the road anymore. The ones in Texas use optic motion sensors mounted on the crossbar the last 10 years. A lot cheaper to replace when they malfunction. Doesn't reduce the rate of malfunction, unfortunately.

Huh?  Most traffic signals around here use loops, even the new ones.

[agentsteel53]

loops here in CA as well.

but that's because no new roads have been built in the last 10 years. 

[Alps]

More jurisdictions are switching to video, but loops are still cheaper. Loops make more sense in warm areas like California or the South and Southwest, because it's the freeze/thaw cycles that buckle the pavement and break the loops. Without that, a loop will easily last for the life of the pavement, and then you just have to replace the loop every N years. All in all, as much as we like to see new, better technology be employed, cost is the 40,000 lb gorilla.

[tradephoric]

The loop is the tried and true method for vehicle detection.  Loops are cheap, simple, and effective and outperform more expensive technologies like video detection, radar, wireless pucks etc..  This is a fact that a lot of jurisdictions don't want to admit because they've been suckered into investing in some advanced vehicle detection technology that doesn't work very well. 

[Alps]

The loop is the tried and true method for vehicle detection.  Loops are cheap, simple, and effective and outperform more expensive technologies like video detection, radar, wireless pucks etc..  This is a fact that a lot of jurisdictions don't want to admit because they've been suckered into investing in some advanced vehicle detection technology that doesn't work very well. 

Tried and true isn't always best. I have looked a lot more than you have into these technologies - I can tell by the way you've phrased this - as part of my work. As I said, each one has its own application. You have to look at all of the circumstances, particularly weather cycles and types of vehicles, to select the appropriate technology. Loops do not "outperform." Don't spread misinformation.

[Central Avenue]

Tried and true isn't always best. I have looked a lot more than you have into these technologies - I can tell by the way you've phrased this - as part of my work. As I said, each one has its own application. You have to look at all of the circumstances, particularly weather cycles and types of vehicles, to select the appropriate technology. Loops do not "outperform." Don't spread misinformation.

You've got me curious now. What might be some scenarios where video detection might be a better option than loops? I've noticed a handful of camera-controlled intersections here in the Columbus area but the vast majority have loops, so I'm wondering what's different about the intersections that get video.

[Alps]

Tried and true isn't always best. I have looked a lot more than you have into these technologies - I can tell by the way you've phrased this - as part of my work. As I said, each one has its own application. You have to look at all of the circumstances, particularly weather cycles and types of vehicles, to select the appropriate technology. Loops do not "outperform." Don't spread misinformation.

You've got me curious now. What might be some scenarios where video detection might be a better option than loops? I've noticed a handful of camera-controlled intersections here in the Columbus area but the vast majority have loops, so I'm wondering what's different about the intersections that get video.

Check my post upstream. Weather can fatigue loops and require them to be replaced every year. Heavy truck traffic, same issue. NJDOT got tired of having to constantly send crews out and maintain loops, so they've gone camera for everything. (That's not necessarily the right approach, but I'm sure they did the math, because they don't have the money to waste.)

[NE2]

Additionally, loops can't detect plastic bikes. Obviously a huge concern.

[Alps]

Additionally, loops can't detect plastic bikes. Obviously a huge concern.

Loops can't detect bicycles in general - too small a magnetic footprint - unless they're tuned to be so sensitive that they pick up other magnetic noise as well. The problem is that loops don't even pick up motorcycles.
One other reason that loops are sometimes not favored - concrete pavement. DOTs don't like to go digging - it's expensive and messes with structural integrity.

[tradephoric]

Let's focus on the effectiveness of video detection.  A 1992 study by MacCarley compared the performance of eight video detection systems and found that detection was inaccurate during transitional lighting periods.  A 2009 study entitled "Evaluation of Video Detection Systems Volume 2 — Effects of Illumination Conditions in the Performance of Video Detection Systems"  came up with the following conclusions (the cameras used included Autoscope, Iteris, and Peek)

During sunny morning, false calls increased in greater proportion in zones where vehicle shadows were more prominent (up to 21% at stop bar and up to 43% at advance zones). Results during the dusk condition followed trends similar to the dawn, but with higher increases in false calls (ranging from no change to about 50% increase); and higher increases in the missed calls observed for one VDS.

After 20 years of research the same problems persist; in this case highly inaccurate counts during transitional lighting conditions.  A 50% false call rate can have crippling effects on an adaptive signals system.  Keep in mind 99% of all traffic signals are basically non-adaptive.  A high false call rate won't have much effect on the operations of a signal that runs a preset timing with limited adaptive operations (ie. simple side-streets gapping or skipping phases).  Basically the shortcomings of any technology can be masked when it isn't being used to its full potential (don't need the security system in your house to work if you never get robbed).

Adaptive signal systems (SCOOTS, SCATS, INSYNC, RHODES ) are the future when it comes to providing efficient signal operations but these systems require accurate counts or they can create more problems than they solve.  A camera that is over-counting (bright sunny morning) can soak up the green time from the main-street even though there may be little or no traffic on the side-street.  Similarly, it's not uncommon for a camera to provide multiple cycles of zero counts (bad filter or Autoscope unit in the cabinet) and may cause the main-street to vote for a minimum green time.  It doesn't look pretty when a major 6-lane boulevard is voting for 10 second greens because the main-street camera isn't picking up any counts for multiple cycles.  When you manage a signal system of 600 signals, half with video detection and half with loops, and the major problems described above only occur at the intersections with cameras it becomes a no-brainer which detection system to choose.

Here's one last finding from a study Perdue did in 2006 that may be impractical for most agencies to adhere to and lead to serious pause when considering the use of video detection.  Intersection with loops can chug along for 10 years with the crew barely opening the cabinet:

The accuracy of all three systems appears to degrade with time and it appeared that a re-calibration was necessary only 4 months after the initial installation by factory representatives.

The misinformation often comes from the mouths of aggressive vendors trying to sell a product.  The fact remains that the studies have shown the same problems that were around with video detection in 1992 are still around today.  Loops outperform video detection when it comes to false calls, missed calls, stuck-on calls, or dropped calls at an intersection.

[mgk920]

This brings up an interesting question. Do any states have a specific time limit after which a motorist is legally allowed to assume a signal is malfunctioning and disobey it?

It's not exactly the same case, but Wisconsin makes an allowance for 2-wheeled vehicles:

Notwithstanding subd. 1., a motorcycle, moped, motor bicycle, or bicycle facing a red signal at an intersection may, after stopping as required under subd. 1. for not less than 45 seconds, proceed cautiously through the intersection before the signal turns green if no other vehicles are present at the intersection to actuate the signal and the operator of the motorcycle, moped, motor bicycle, or bicycle reasonably believes the signal is vehicle actuated. The operator of a motorcycle, moped, motor bicycle, or bicycle proceeding through a red signal under this subdivision shall yield the right-of-way to any vehicular traffic, pedestrian, bicyclist, or rider of an electric personal assistive mobility device proceeding through a green signal at the intersection or lawfully within a crosswalk or using the intersection. This subdivision does not affect any authorization for a bicyclist under subd. 2.

Other than that, I can't seem to find anything related to a malfunctioning signal in our statutes. One would hope to assume common sense would prevail, Read: Check for conflicting traffic and go when the way is clear.

I recall a discussion on that several years ago and from what I gathered, that Wisconsin law applied to all motor vehicles.

I wonder if it was changed at some point.

Mike

[tradephoric]

Weather can fatigue loops and require them to be replaced every year. Heavy truck traffic, same issue.

To suggest loops will only last a year in states that experience cycles of frost heave or states with high weight limits on trucks is highly inaccurate.  Loops can easily last 15 years with little maintenance (replacing a loop rack, etc). even in harsh weather states. 

NJDOT got tired of having to constantly send crews out and maintain loops, so they've gone camera for everything. (That's not necessarily the right approach, but I'm sure they did the math, because they don't have the money to waste.)

You've never heard the phrase "wasteful government spending" before?

[Alps]

Tradephoric, can you come out and say which company you work for that manufactures loops? Because this is getting very tiresome. For the record, it's better to have false calls - which video systems are designed with intentionally - than to miss calls, which loops have been known to do (broken connection, for example). Better to have the side street turn green too much than never.

[tradephoric]

For the record, it's better to have false calls - which video systems are designed with intentionally - than to miss calls, which loops have been known to do (broken connection, for example).

That's inaccurate.  A detector unit provides a continuous call in the controller if the loop circuit is incomplete (broken connection).

There's undoubtedly problems with both loops and video.  I may have been a bit harsh with video detection in my initial post but I'm not the only guy who has seen the problems that exist with video detection.  This is an abstract that was written by the Texas Transportation Institute:

Due to the well-documented problems associated with inductive loops, most jurisdictions have replaced many intersection loops with video image vehicle detection systems (VIVDS). While VIVDS have overcome some of the problems with loops such as traffic disruption and pavement degradation, they have not been as accurate as originally anticipated. The object of this project is to conduct evaluations of alternative detector technologies for application into the state's traffic signal systems. The research will include investigating the available detectors that could replace loops or VIVDS through a literature search and agency contacts, followed by field and/or laboratory investigations of promising technologies.  (FHWA/TX-09/0-5845-1)

Video detection has always had potential but has never been able to achieve the accuracy needed for most agencies to be a viable long-term solution.