[Tagging] Micro mapping traffic signals

Fernando Trebien fernando.trebien at gmail.com
Wed Aug 28 14:57:06 UTC 2013

I've been inclined to combining traffic lights
(highway=traffic_lights) and pedestrian crossings
(crossing=traffic_lights), since they usually form a single "stopping"
structure. The traffic light node then represents the "stop position"
of the cars, even if the light itself is placed in the middle or at
the other end of the intersection. This works well when mapping
cycleways and sidewalks as ways and also works well for pedestrian
routing, while keeping the map somewhat simple. As a bonus, it renders
nicely on both Mapnik and OpenCycleMap. Example:

However, routing would double count traffic lights in two-way roads
(as in Kytömaa's counting), though guessing the direction by proximity
to the intersection should be accurate here in about 99% of the cases,
as would be for stop signs.

This may vary per country, but I wonder how mappers who do not work
for the local transit agency would know for sure which traffic lights
are synchronized. Assuming that mappers would survey the area for
traffic light timings, they would still get different results if they
survey at different times of the day (as Pieren mentioned). So the
"wave" information is not "on the ground" (a common mapping
recommendation), and that is likely to lead to edit wars.

At least here in Brazil (so this may not apply elsewhere), I've
empirically observed that the following rules work almost all the time
(and thus may be part of some router's logic):
- I wait the longest at a traffic light when crossing or changing to a
road of a higher class (normally from a secondary to a primary road,
where most lights are found), and then get quite little green time
- this also applies somehow to "stop" signs, but here the wait is
longer and the opportunity is shorter due to more intense traffic on
the higher-class road
- traffic lights at crossings of same-class roads seem to give about
50% of time to either direction (it may vary up to about 40% to one
direction and 60% to the other)
- having entered that road, I'd usually get a red light only at the
first traffic light, and from then on all traffic lights (or most of
them) would be synchronized as long as I follow a along ways of same
class and keep crossing ways of lower class

This, however, follows from my personal and unofficial "classification
proposal" for Brazil (which still divides opinions here). Brazilian
prefectures only inform which streets are arterials, but not which are
collectors. My proposed "algorithm" then classifies arteries as
primaries and all other urban roads as residential at first, then
promotes to tertiary those with right of way (treating them as
collector roads), and to secondary those that have right of way over
tertiaries. In uncontrolled intersections, right of way is established
by the presence of stop signs. If none are present, then no way gets
promoted. In controlled intersections, right of way is determined by
the direction that gets more green time. Some have complained that my
method leads to a high density of tertiaries, and I agree, though I
still believe this is merely aesthetic. It works quite well in areas
laid out as as grids. It should also work well on fused grids. It does
not work so well in irregular areas. Here's a test case, applied all
over my own city:

On Wed, Aug 28, 2013 at 1:16 PM, Pieren <pieren3 at gmail.com> wrote:
> On Wed, Aug 28, 2013 at 3:01 PM, Martin Koppenhoefer
> <dieterdreist at gmail.com> wrote:
>> (to reduce traffic in certain areas). Routing algorithms would have to know
>> all phases of the traffic lights for a perfect result.
> About the "green wave", it could be modelized with another relation
> (or super-relation) grouping of traffic signals and the direction of
> the wave. After all, this is also what is happening in the real world
> where such wave is only possible when the concerned traffic lights are
> synchronised by one machine.
> Defining different "phases" on day time or night time is outside of my
> proposal for relations.
> But another point is traffic signals where a bicycle can ignore red
> light stop when he is turning to right (right-hand traffic). This is
> allowed in some coutries, sometimes only with a specific road sign. I
> don't know how it is tagged currently.
> Pieren
> _______________________________________________
> Tagging mailing list
> Tagging at openstreetmap.org
> http://lists.openstreetmap.org/listinfo/tagging

Fernando Trebien
+55 (51) 9962-5409

"The speed of computer chips doubles every 18 months." (Moore's law)
"The speed of software halves every 18 months." (Gates' law)

More information about the Tagging mailing list