Episode 8 – Signal Detection & Operation

Topics: Signal Detection and Actuation, Inductance Loop Detectors, Video Detection

Talking Traffic Episode 8 – Signal Detection & Operation

Hello and welcome to Episode 8 of Talking Traffic. My name is Bill Ruhsam and I am the creator of this podcast, and its sister website, Talking Traffic.org. Today is Wednesday, January 15th, 2008, the first episode of the new year! Today’s episode is about Signal Detectors and how they operate. The various gizmos that let a traffic signal see that you are sitting there in your car, wanting to make a left turn, or whatever. There is a great deal to be said on this topic but I’d like to cover a couple of definitions first, so that we’re all on the same page.

Way back in the first episode of Talking Traffic, I explained the definition of the word “Actuation” as it refers to traffic signals. Repeating this, to actuate a signal is to cause it to switch to another signal phase. Which means, if you arrive at a red light, to actuate is to summon the green for your direction. The most common use of actuation is for left-turn signals, where the signal will not give a left turn arrow unless a vehicle arrives and actuates that phase.

This brings us back to the question of how do we actuate a signal? How does a signal detect you? There are several common methods. They boil down to Inductive detection, Video detection, and other various sorts of detection.

First let’s talk about the inductive detector. I’m sure you’ve seen the rectangular or square cuts in the pavement, either on the approach to a traffic signal or at the stop bar. These cuts have wires in them that run back to the signal controller. They form an inductive loop which senses the presence of metal through an electrical property called inductance. The detailed explanation of inductance is tricky, but what it means on a day to day basis is that the detector loop creates a magnetic field above the pavement which has a certain inductance value. When a lump of metal such as a car, motorcycle, or bicycle (yes, bicycle) sits inside that magnetic field, the inductance value changes, and the signal controller can sense that change. So, to activate an inductance loop, you only have to be on top of it with metal. Not even magnetic metal, so aluminum cars will be fine.

You will see these inductance loops cut into the pavement in several different ways, with numerous variations. They can be broken down into two very broad general categories: Presence detectors and Pulse detectors. The purpose of the Presence detector is to sense when a vehicle is sitting on top of it. These you will see right at the stop bar. Pulse detectors are used for in-motion vehicles, to signal that one has passed over the detector without stopping. Presence detectors are (again, generally) 6 feet wide by 40 foot long rectangles, and pulse detectors tend to be 6 by 6 square. There are nitty gritty electronic technical differences in all of these, but the details don’t matter that much to the public. If you’d like to see a good description of how inductive loop detectors work and are applied, see the show notes for a link to a PowerPoint presentation by engineer Bob Shanteau in California. The presentation concerns how to detect both Bicycles and Motor Vehicles using the same detector, but it does an excellent job explaining how an inductive loop detector works.

The next type of detector that we are seeing more and more of every day is the video detector. A small video camera is mounted on top of the traffic signal pole, or on the arm that holds the signal lights. This camera is aimed at the stop bar where cars will be waiting. Contrary to popular belief, these cameras are not used for video monitoring. They don’t have the resolution of a security camera, nor do they rotate or tilt. They must keep pointing at the same location to function. How they work, is by detecting changes in the contrast of their detection zone. Imagine that you are looking at a video screen with the image of a traffic signal from the vantage point of the camera. You are looking down at the street, at the stop bar, probably with vehicles facing you ,waiting for the green light. A signal technician uses a computer to draw areas on that video image that the computer will then monitor for changes in pixel contrast. For example, at a left turn bay, if there are no vehicles there, the camera will see some dark pavement, maybe an arrow striped on the ground. When a vehicle pulls up to the stop bar, suddenly the pixels that are within the detection zone change in value; no longer does the camera see dark pavement and light striping, now it sees bright metal and glass, or black paint, or whatever. The point is that the view changed, and the camera can detect it.

As I just mentioned, these detection zones are drawn using a computer onto the image that the camera views. A technician works in the signal controller cabinet with a laptop placing the loops where they are needed to operate the signal. These loops are variable in size, length, and location; they can be moved or resized at the touch of a button. That makes video detection quite cool from a technical aspect.

The inductive and video detectors both have pros and cons. Inductive detectors require cuts to be made in the pavement and wire to be installed . They are not movable. If someone decides that the detectors should have been somewhere else, new cuts and new wires must be installed. Also, the wires are easily damaged or severed. You’d be surprised how much the pavement at an intersection moves over its lifetime, and it only takes an inch or so to sever the loop. A periodic maintenance and replacement program is essential with loop detectors to keep signals operating correctly. On the plus side, Loop detectors are relatively cheap. They include saw cuts, wire, and epoxy, That’s it! A good contractor can do a whole intersection very quickly.

Video detectors on the other hand are not cheap. You have a host of necessary equipment, including the cameras, the communication cables, the computer to run the detectors, and the interface with the signal controller cabinet. However, once installed, they can be altered at will by a technician. The video detection areas are not limited by the types of geometries that are forced on inductive loops by the electronic necessities of inductance. Unfortunately, video is problematic because of what it is; visual detection of changing pixel contrast. The rising and setting sun can cause the video detector to go into what is called recall (more about that in a moment). Snow, fog, rain, and other inclement weather can cause the detector to stop working correctly. (if you lived where I lived in Texas, then Dust can be the issue). Fouled lenses, or plain ‘ol broken cameras will cease the correct operation of the signal. Also, for those pulse detectors I mentioned earlier, you need to set those back from the intersection quite a ways. Video cameras up at the intersection can’t see that far because their angle of vision will be blocked by trucks sitting at the stop bar, which would cause spurious detections. This circumstance would need another set of cameras on poles set back from the stop bar by the appropriate distance. More cost, there. So there are issues that the agency that maintains the signals have to contend with when deciding which type of detector to go with, inductance or video. Most of the issues revolve around staffing, maintenance, and budgets: Is there staff available to do year-round loop maintenance? Can they afford the up front costs of the video detectors? What are the lifecycle costs of both systems? it gets complicated.

The “other” category of signal detectors encompasses a bunch of different technologies including microwave detection, ultrasound, mechanical, magnetic pole sensors, and other more esoteric items. These aren’t seen too often nowadays because they tend to be expensive, or intended for specialized applications. The only one of those we’ll talk about much unless I get a lot of requests is the pedestrian pushbutton, and that’s for another episode.

The last thing I want to discuss is how those signal detectors affect traffic signal operations when they break. Remember I mentioned the word “recall.” Again, there is nitty gritty detail if you talk to signal engineers about the word, but what it means to the common driver is, “this darn detector is broken!” There are two ways a signal detector can break, either it sends no signals at all, or it sends a signal constantly. Either way, the signal controller is programmed to go into recall mode, which means that it will always serve the phase that the broken detector is calling for. If you’re in the middle of nowhere, with no one around and the left turn phase that no one is sitting for comes up, the signal detector is probably broken. Of course the other option is that there are no detectors and the signal is running on a fixed time cycle, serving every phase every cycle.

Proper maintenance on signal detectors is important if you don’t want to have upset drivers. If you constantly sit at a red light because a broken detector is calling up the side street green, you’re going to write to your legislator. Trust me, engineers do not like it when legislators call to ask why something is going wrong.

Now, I hear that voice out there saying, “But I’ve sat at a signal before without getting the green light.” I’ve done that, too. My first thought is always, “am I sitting on the detector?” If you pull up too far, or if the signal detector isn’t placed correctly, you might not be tripping it. Some detectors are programmed to ignore a signal if it goes away after a few seconds, this enables people who are making right turns on red to avoid calling up that phase. However, if you pull up too far, you may not have indicated to the signal controller that you’re there, and waiting. My second thought on the other hand is, “why did they program this signal to skip my phase every other cycle?” Sometimes that happens, but it will usually on happen on very busy arterials during rush times. If that’s not the case then my third thought is, “What is the phone number for the local traffic operations office?” If you see a signal not working, or a signal light that is out, please call the local operations office. If no one tells them things are broken, they can’t fix it.

Thanks for listening to another signalized episode of Talking Traffic. Next time, we’ll do a topic about other things than signals. Something more interesting or weird. I’ll find something funny, even. I guess you’ll have to download to find out. If you have any comments or questions, please email me at bill at talkingtraffic.org or leave a note at www.talkingtraffic.org.

This episode is released under a creative commons 3.0 license. You are specially encouraged to share the episode, and blog about it, but please don’t change it or sell it. My cat depends upon this podcast for her food. Ok, not really, but with the internet, it’s very easy to think anything you download is yours for the taking and changing. I encourage you to become familiar with Copyright law, Intellectual Property Principals, and the Creative Commons. That is my public service announcement of the month, and now, let’s end the podcast with the full version of Five Star Fall’s, Mercurial Girl. I love the beat on this song…

PowerPoint Presentation by Bob Shanteau: Detecting Bicycles and Motor Vehicles Using the Same Loop Detector
Music: Five Star Fall

This entry was posted in podcast, signals. Bookmark the permalink.

6 Responses to Episode 8 – Signal Detection & Operation

  1. Pingback: The Evil Eyebrow » Blog Archive » Podcasting Music Angst

  2. James Cronen says:

    Great episode, Bill!

    Is there any considerable numbers of drivers that just get fed-up with traffic lights that are too long? If so, what do traffic engineers consider the longest acceptable wait?

    I remember growing up in New Jersey I once sat at a light for seven minutes one late night before I finally got the green. Now that I think about it, I’m pretty surprised that I even waited that long without running it.

  3. Bill Ruhsam says:

    There are plenty of statistics that deal with red-light running, but I don’t know them off the top of my head. I’ll have to look into that.

    For total cycle lengths, it’s common to go as long as 120 seconds, but you’ll only wait a portion of that for the green unless you just miss the signal. In high traffic areas along major corridors, the cycle length can go up from there. I know of one set of signals around here operating on a 210 second cycle length, but that is unusual.

    A red-light running/enforcement episode is a good idea. I’ll put it in the stack.

  4. Annie says:

    I’ll be waiting for that pedestrian call-button episode. :) I still firmly believe that all traffic, pedestrian and vehicular, would move through the intersection much more efficiently if the pedestrians waited for there protected light (referring to the email I sent you some months ago).

  5. Bill Ruhsam says:

    Annie: You need to send me the google maps coordinates of that intersection so I can look at it.

  6. Nick says:

    I collect traffic signals of various vintages as a hobby. Mine are video actuated with the processor and all the electronics in the cabinet.

    Here’s a video of them: http://www.youtube.com/watch?v=sdQwsQjz98g

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>