Do pedestrian buttons actually work?

When people find out I have a working knowledge of traffic signals, this is often the first question they ask.  “When I press the button, does it actually do anything to the signal timing?”  The short answer is: sometimes.  Because there’s more to pedestrian buttons than just getting a walk light to cross the street.

There’s two parts t0 the answer here, which are:
1. There are some pedestrian crossings where someone needs to press a button for the walk light to be displayed, and other crossings where there is no need for anyone to press a button,
2. Pedestrian buttons are being installed at all pedestrian crossings, whether or not those buttons actually have an effect on signal timing.

Why doesn’t the walk light just come on all the time?

At a typical intersection between a main street and a side street, the pedestrian signal along the main street defaults to walk, so it doesn’t matter whether or not anyone presses the button. But to get a walk signal along the side street, you actually need to press the button. If there are cars waiting and nobody presses the pedestrian button, the vehicle signal will display Green but the pedestrian signal will remain in Don’t Walk.

If the walk signal came on every time the light turned green, it could cause unnecessary delay to perpendicular traffic in the situation when there are only a few cars and no pedestrians.  For example, a pedestrian crossing across a 7-lane road might require a minimum of 28 seconds of green (7 seconds of Walk and 21 seconds of Flashing Don’t Walk).   But in the absence of a pedestrian call, the green duration for side streets can respond in real-time to the actual amount of vehicle demand, so if there are only a couple cars waiting, the green light could be as short as 7 seconds.  Varying the green duration based on side street vehicle demand means that people travelling along the main street aren’t kept waiting any longer than needed*.

It’s for this same reason that often pedestrian crossings along a main road will display Walk regardless of whether anyone presses the button.  Due to the anticipated volume of vehicle traffic, the signal is going to provide more than enough time to walk across the side street anyway, so the pedestrian signal might as well display Walk even if there wasn’t anyone waiting to cross at the start of the green.  Displaying Walk as much as possible avoids any unnecessary delay for people who show up after the light has already changed to green.

Why are all crossings getting pedestrian buttons?

Eventually, all pedestrian crossings in the City of Toronto will have pedestrian buttons, regardless of whether the pedestrian button would have any effect on signal timing.  The same is also true in many other jurisdictions across the province.  This is because buttons are the way of activating the Accessible Pedestrian Signal (APS) audible tones, which are gradually being installed at all traffic signals.

Originally, APS was on whenever the walk light was on, so it didn’t need any buttons to operate.  But after a number of noise complaints from nearby residents, the current system was introduced, where the APS sounds would only be played when someone presses and holds the pedestrian button for 3 seconds (at which point the button makes a tick noise to indicate the APS call was received).

Making sense of pedestrian buttons

The problem with having buttons that don’t affect signal timing is that when some people discover that there are pedestrian buttons which have no effect on signal timing, they come to the conclusion that all pedestrian buttons are placebos.  This is a problem, because there actually are crossings where one does need to press the button to get a walk light.

The solution to this problem is very simple.  Each pedestrian button should indicate whether one needs to press it to get a walk signal, or if it’s only there to call the APS sounds.  Variations of this system are implemented in a number of jurisdictions in Ontario.  In the City of Ottawa, buttons which actually call the pedestrian phase have a sign saying “press this button”, while buttons which only call APS sounds have no sign.

In Ottawa, the buttons without signs are only there to activate APS sounds

In Ottawa, the buttons without signs are only there to activate APS sounds

But I found that system to not be sufficiently reassuring.  Upon encountering a button with no sign, I wouldn’t be certain if meant that I didn’t need to press the button or if the sign had simply fallen off.  So I always pressed the button to be sure.

My favourite solution is the one used in Waterloo Region, where buttons which only exist to activate APS sounds have the symbol of a person with a white cane.

This button only activates the APS sounds

“This button only activates the APS sounds”

This button calls the pedestrian phase as well as the APS sounds

“You need to push this button to get a Walk light”

This labeling system is a clean and simple way of indicating what pedestrian buttons do (or don’t do), putting an end to the myth of the placebo pedestrian button.

*Endnote: Actually with the typical location of side street vehicle detection loops, there tends to be about 3 seconds of unnecessary green after the last car goes through, but this is still a lot less than the dozens of seconds that could be wasted if the walk sign came on when no pedestrian wants to cross the street.

This entry was posted in Traffic Signals. Bookmark the permalink.

12 Responses to Do pedestrian buttons actually work?

  1. This is of course all true, I am still incredibly frustrated by having to hit pedestrian beg buttons, and generally unwilling to stand a whole green cycle when I arrive just as the light turns green and thus cannot push the beg button. This is frustrating on a road like Fischer-Hallman where there are few pedestrians and it is an understandable situation–although there is no reason that pressing the ped button shouldn’t turn on the crossing light immediately, as the light will also be extended if there are more cars crossing the underground loops to signal the light (provided there is time in the phase). The fact that cars can get extra time (in fact, often more than enough time for peds to cross) but pedestrians cannot is simply insulting. At an intersection like West and Victoria, where I live and thousands of pedestrians live every day, it’s just plain wrong that there is a beg button. But if you want a even more extreme example, the intersection of King and Willis, which I have no doubt sees more pedestrians than vehicles, also required a beg button.


    • The problem is with the pretimed signals. If you were to walk up to a green light that was fully actuated and press the walk button, it should extend the green and give you a walk phase.

      Liked by 1 person

      • It doesn’t need to be fully-actuated for that to be true, it can be the case with any actuated phase. For an intersection to be fully-actuated, every phase needs to be actuated. But intersections can still be very responsive while being semi-actuated, which means that all but one phase (usually the main street) is actuated. For example, the signal at the start of the video is uncoordinated Semi-Actuated Type 2 (SA-2), and you’ll notice that the walk light comes on the instant I press the button. There’s a couple things I would change about the programming there to reduce pedestrian delay, but in general it’s pretty pedestrian-friendly even though all the signals default to Don’t Walk. Because it is uncoordinated, the cycle can be as short as possible to keep wait times short.


    • I guess you’re saying that pedestrians shouldn’t have to ‘beg’ to cross the street, but the fact is that if you want to have traffic signals operating efficiently, it helps to know who wants to go where. And buttons are the best way of doing that for pedestrians. There’s no way a video/optical detector would be able to identify whether any individual intends to cross the street, or which direction they’re waiting for. Avoiding unnecessary green time helps not only motor traffic but pedestrian traffic as well. If a walk light is unnecessarily triggered across a main road, not only do the cars need to wait, but so do the pedestrians.

      What matters is not whether you need to press the button, but rather how quickly the signal responds once you do. There is a wide variety of practices and objectives across Ontario jurisdictions, with a wide range priority for pedestrians. Some jurisdictions, such as Toronto, place a very high importance on pedestrian wait times even at the expense of motor traffic flow (though the current mayor is trying to reverse this priority). Some other jurisdictions (I’ll refrain from pointing fingers) will make pedestrians wait if there’s even the slightest chance of impacting motor traffic.

      There’s no question that if the light is green and there’s still time in the phase to serve a pedestrian call, the walk light should come on immediately when you press the button. The catch is that there are few intersections where that could ever occur. At a minor street, the green duration required for a pedestrian to cross the main road is typically way more than the green that would be required for the volume of motor traffic. So the phase duration is equal to the minimum pedestrian time and therefore if you miss the start of green, you miss the entire cycle. This is very harsh, and the way of being a bit more forgiving is called the permissive period, which I’ll cover in a later post.

      In some jurisdictions, all signals are Semi-Actuated 2 (where the side street walk light is only displayed when someone presses the button). That’s just silly, because there will be many instances where the walk light could have been displayed with no impact to other traffic. But in more pedestrian-friendly cities, the decision about mode of control (ranging from fixed-time on one extreme to fully actuated on the other) takes into account the volume of traffic and the number of pedestrians. If either of the following are true, the walk light will come up every time the green does:
      1. There is enough pedestrian traffic that the walk light would tend to be called the vast majority of cycles anyway, or
      2. The volume of traffic is enough that the green is usually long enough for a pedestrian to cross anyway.


      • First of all, I think its important to put some context here. On Fischer-Hallman, yes, rarely is there a pedestrian waiting to cross, and it makes sense to provide more time to automobiles that are actually using the main through street.

        However, on Willis Way at the Waterloo Uptown Pedestrian Square, it is ridiculous to make pedestrians “beg” to cross, it should simply turn on the ped signal every phase.

        Secondly, there is some misunderstanding or confusion here about what I am referring too. Traffic signals on side streets which are only triggered by detection loops/buttons and not timers, aren’t green for a fixed time. The light will stay green so long as the detection loop detects more cars coming (up to a maximum time allotted), and when there are no more cars or the maximum time has elapsed, it will turn red. Because the signal doesn’t know how long it will stay green in advance, it won’t turn on the pedestrian signal since it might only be one car waiting and the green phase will be short.

        However, if as a pedestrian, I arrive at the intersection as it turns green, and press the button, the light green time won’t be extended for me, even though there may be sufficient time within the maximum time the signal can be green for the pedestrian phase to go. Worse yet, since many cars may be lined up, the light might stay green for far longer than a pedestrian phase anyway.

        You can see how this is by far an unreasonable situation. This isn’t an edge case either, it occurs for me approximately 1/3 of the time I cross the street near my house.

        When it comes right down too it, delay is what matters most, but making me wait and face a green light for 30 seconds or more, before waiting for a red, just to get green again, is a problem too.


      • Based on your description, King & Willis Way meets the warrant of “lots of pedestrians” so it would likely be changed from SA-2 to SAP (SA-1) if it were in a jurisdiction that prioritizes pedestrians. So either Waterloo Region doesn’t have a pedestrian-oriented warrant system for mode of control, or they do and this signal hasn’t been audited recently. Either way, something needs to change.

        I am clear on what you’re referring to regarding extending actuated greens. You said that if there’s enough time left in the phase to serve the Walk+FDW, it should do so immediately. And I said that I agree. This operation would be making use of a permissive period. For example, if the maximum green is 5 seconds longer than the minimum pedestrian time, then there would be a 5 second permissive period at the start of the green where pedestrian calls would be served immediately.
        The problem is that usually the maximum green is equal to the minimum pedestrian phase time. So after the light turns green there is no longer enough time to serve the pedestrian phase.

        Liked by 1 person

      • I am unfamiliar with the terms you are using. Region of Waterloo traffic signals definitely have the ability to operate in a mode where the pedestrian phase is automatic, and no button needs to be pressed, as this happens at many intersections between two major roads. I’m not sure if this is what you’re referring to.

        As for green phases, I cannot speak to most intersections, but for the several I have observed, it is definitely not the case that the pedestrian phase is equal the the maximum green phase. This is also apparent in the situation where the green phase is extended past the pedestrian phase (which can also happen). In this situation you will see the pedestrian counter count down timer reach zero, and then the light remain green. I find this annoying as both a pedestrian and a driver.

        Further, the pedestrian phase need not be as long as it is. The phase usually involves a walk sign for several seconds, then a count down. The walk phase need only be 1-2 seconds long as a minimum. The point of leaving the walk phase long is to provide pedestrians the maximum time to cross on lights that will be green for a while, but it is not necessary for it to be long. Which means the pedestrian phase at the intersection I live near could be as short as 8 seconds, meanwhile, I will wait, and I have timed it, at a green for over 30 seconds.


      • Sorry, I assumed you read the article that Sylvan linked, which defines the types of mode of control. For semi-actuated signals (which default to one phase, and serve the rest only based on demand) the two main categories are:
        SA-1, or SAP, where the walk light will come on any time the green does, and
        SA-2, or SA where the walk light only comes on when someone presses the button.

        In the example you give of a semi-actuated intersection where the side street phase has more than the minimum pedestrian time, I see a few options to improve level of service to pedestrians.
        1. As you suggested, maintain SA-2 operation but use a permissive period so that it actually serves pedestrian calls whenever there’s enough time in the phase,
        2. Serve a minimum pedestrian call every time the light turns green (i.e SAP)
        3. Serve a vehicle-extendable pedestrian call every time the light turns green
        4. Convert to fixed-time.

        These happen to be ordered from lowest main-street delay (#1) to highest main-street delay (#4). And correspondingly, highest side street delay (#1) to lowest side street delay (#4).

        The minimum pedestrian time is a certain minimum walk followed by a flashing don’t walk clearance based on the crossing distance. Waterloo Region apparently uses a 10-second minimum Walk followed by a Flashing Don’t Walk (FDW) based on crossing at 1.25 m/s. So for example at Fischer Hallman and Thorndale, the crossing distance is 27.5m, so the minimum pedestrian time is 10s Walk + 22s FDW (32 seconds). 10 seconds does indeed seem to be longer than necessary, given that the Walk signal merely needs to give you enough time to react and start crossing the street. The City of Toronto, for example, uses a 7-second minimum Walk. Typically, calculations use a 2-second perception-reaction time, so a 4-second minimum walk should be adequate as long as people are paying attention.


  2. Do you remember Mark Wagenbuur’s blog post about traffic lights? The pedestrians are affected in much the same way that bicycles are, they just don`t have advance detection.

    It is literally possible in Den Bosch to have a complete cycle with 4 phases (independent turn phases) of just 28 seconds for motorized and bicycle traffic, and pedestrians have a bit more, assuming a 9 metre wide road to cross at 7.5 seconds per arm, the complete cycle can happen in just 37 seconds at most. Full actuation matters a lot.

    But their practice of keeping junctions as small as possible, designing it so that yield signs alone could in theory govern the intersection and using the protected intersection has another benefit, it allows the signal to turn off at night, say from 11:30-5:30 or so and remain safe. Many traffic lights in Utrecht have been just switched off and bagged up. It’s something I’d like to see as well.


  3. Sylvan says:

    The design of SA2 signals in Ontario is unnecessarily pedestrian-hostile. I read a great comment on Spacing a couple years back identifying a small change we could import from BC to improve pedestrian wait times when crossing major arterials at low-traffic cross streets:

    I also wonder whether the same camera tech that is used to detect cars at intersections (in lieu of induction loops) could be used to automatically detect pedestrians waiting to cross at SA2 signals, eliminating the need to use the beg button.


    • This is a great comment, thanks for the link.


    • Some of the “Vancouver” vs “Toronto” comparisons in that comment are unfair because both methods are used in both places. Nevertheless, some of these methods are indeed ways to reduce pedestrian delay.
      1) This item describes the difference between co-ordinated and uncoordinated (“free”) operation. In coordinated mode there is a constant fixed cycle length, which allows the signal to stay in sync with other nearby signals, opening up the possibility of arrangements like green waves. With free operation, the signal operates independently so it responds solely to the demand at or approaching the intersection at the moment. Free operation is better in every way, except that it doesn’t have a fixed offset relative to adjacent signals. In Toronto most signals operate in coordinated mode, but there are also some signals that operate free.
      2) Toronto rarely uses bicycle pushbuttons. That is indeed annoying because although the in-pavement detectors can already detect bikes if they’re calibrated correctly, not all of them are, so it’s nice to have a reassurance button.
      3) This item talks about walk recycling, but mixes up a bunch of different things. I’ll try to sort them out.
      In uncoordinated (free) mode in Toronto, the signal will stay in Green/Walk for the main street indefinitely until someone requests some other movement. The only reason a free signal would revert directly back to Walk at the end of the countdown is if a car shows up but then turns right on red before the countdown reaches zero.
      In coordinated mode in Toronto, the main street countdown occurs every cycle regardless of demand. Far from being an anti-pedestrian method, this actually reduces wait times for pedestrians. In the absence of a permissive period, a coordinated cycle will only have one instant per cycle when the main street signal is permitted to change from green to yellow (the “force-off point”). As the signal approaches this point it will always start the main street pedestrian countdown, even if no one requested the side street phase. That way, anyone who shows up along the side street during the Flashing Don’t Walk can still be served this cycle. Otherwise, there wouldn’t be enough time for the main street to count down to zero before the force-off point, and those people would be forced to wait until the following cycle. In coordinated mode, forcing the main street pedestrian countdown every cycle regardless of demand reduces the maximum pedestrian delay along the side street by the duration of the main street Flashing Don’t Walk, a benefit which far outweighs the minimal increase in delay for pedestrians along the main street.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s