Cycleway "lane" or "track" for Protected Bike Lanes?

In OSM we distinguish road-related cycleways mainly between “lane” and “track”. One could say, roughly defined, that “lane” usually refers to a cycleway directly on the carriageway, and “track” to everything off the carriageway.

But our definition is differentiated and uses the “physical” separation of the cycleway:

  • “A cycle lane is bicycle infrastructure that is an inherent part of the road, but set aside for the exclusive use of bicycles, whilst being separated only by paint or other markings, and without a physical separation from vehicles.”
  • " A cycle track is separated from the road by curbs, parking lots, grass verges, trees or another physical barrier, but is running parallel to and next to the road."

A cycle lane like in the picture below, which is not only separated by markings but also by e.g. posts or bollards, might therefore be a “track”. I would like to get a sense of the community’s opinion on this, to sharpen the definitions for this case.

The example picture is one of our “separation” schema examples, and I think it illustrates the issue very well:

  • Which value would you choose?
  • “lane”
  • “track”

0 voters

In the case the cycleway on the image above shows a “track”, I wonder whether we see a “track” in the middle of the road and a “lane” on the right street side in this image:

So what do you think?

At least in Austria there are dashed lines and solid lines. The first may also be driven on by motor vehicles, since the width of the road is usually not sufficient for two vehicles to cross. On solid lines, vehicles are not allowed to drive on them.
Both I would tag as cycleway=lane, but the definition doesn’t quite fit with the dashed line in my eyes.

I’ve voted for “track” in the picture because of the other physical barrier statement.

According to the respective description, I would tag the second image as cycleway=lane, since there is only a barrier from one side. The vehicle that took the picture could (accidentally or willfully) drive onto the bike lane. This would not be possible with the first picture, because there are bollards on the left and a curb on the right).

Short answer: It is called a protected bike lane.

Long answer:
I’d use the same definition as in the street parking proposal: If it is on the road surface and could easily be transformed into a car traffic lane, it is a lane. In your examples, only a few bollards and some paint would have to be removed, and the cycleway is gone.

I think the difference between protected bike lanes and only paint bike lanes should be recorded with different tags. If I am not mistaken, you already have a proposal for that in your back pocket.

1 Like

I see it the same way, but that would require an adjustment of the OSM definition. Some would argue, a “redefinition” that isn’t easy.

The schema you mentioned is the separation schema (also mentioned in post above).

P.S. Sorry, just realised that I failed with the voting headline (that is a third answer option now :slight_smile: ) but if I correct that now, the previous votes would be lost.

1 Like


for me, the first example is a track because turning left (or changing onto a lane for motorized traffic) is made almost as difficult as crossing a kerb. You have to slow down and find a gap between the bolloards. (Turning left is not relevant if the two carriageways are separated)

The second example is a lane because the separation from the turn right lane is just a painted line.

Please mind that there are more types of protection for bike lanes than separation by bollards. Frankfurt am Main frequently uses “Leitboys” and “Leitschwellen”.

Best regards



Maybe this is stupid, but for me it doesn’t make a difference as long as it is on the road surface. Even the “Leitschwellen” seem to just be attached to the surface of the carriageway and easily removable if wanted.

1 Like

Wikipedia is of a different opinion:

A cycle track , separated bike lane [1] or protected bike lane (sometimes historically[2] referred to as a sidepath ) is an exclusive bikeway that has elements of a separated path and on-road bike lane. A cycle track is located within or next to the roadway, but is made distinct from both the sidewalk and general purpose roadway by vertical barriers or elevation differences.

For bikeways separated from other traffic only by painted markings, see bike lane.

For me, some flimsy pieces of plastic are not a significant enough physical barrier to change a lane into a track. If instead there was a metal or concrete wall or fence, then I would use track.

1 Like

In your sense, the only requirement is how easy to dismantle or change something to have the same road surface again?

So this would be a lane because the black bars could be removed pretty easily and it’s on the same road surface? Whereby anything larger than a city SUV could simply drive over it.

This one too, since all you’d have to do is remove the wooden beams and it would be all on the same road surface again, right? All you would need is a chainsaw.

Again, by your definition this would be a lane because you could just remove the sand and fill it in with some asphalt to have it on the same road surface? Depending on the bike or car, you wouldn’t even notice the pavement changing briefly.

This one is probably a bit more effort, but the two curbs and the bit of greenery in between can be removed, therefore this would be a lane again right? The curbs are apparently only on a small concrete foundation and could be quickly removed with a larger hammer to achieve the same road surface.

Last but not least, when the concrete walls are picked up by a truck, it is on the same road surface. So that a lane?

Disclaimer: This is intentionally written in a clearly sarcastic manner and is not meant to come across as evil or personally attacking. :slight_smile:

My 2 cents:
I argue more from the point of if the ‘lane’ can be entered without leaving the road (surface). In other words, if they are really separated, or if there are just some (platic) bollards indicating the cycle lane.

Yes, because it can easily be crossed (by almost anything) while remaining on the asphalt – i.e. not separated.

No, because it cannot (reasonably) be crossed (at least not by the two main vehicles here; cars and bicycles). Thus, they are separated.

No, and not even by the definition of @Discostu36, as it clearly is a different way (separated by a different (natural) surface). Unless you are telling me that there is asphalt underneath that sand.

Now you are just trolling – this is clearly not what was meant.

When picked up, then yes. Until then, these can’t be (reasonably) crossed, thus no – these are seperated.

Agreed on all counts. A few plastic bollards or bumps won’t stop a car from simply traversing into the cycle “zone”. It is still part of the main highway (road). A separated track I think needs substantial road furniture (kerb, fence, railings) between the bike track and the rest of the highway.

(Of course in the case of an accident, or serious intent, a kerb or fence probably won’t stop a vehicle from crossing into the cycle “zone” but that’s extreme).

An off-topic comment...

I don’t particularly like cycleway=track owing to the potential conflation of the value with highway=track but that’s moot at this point.

Sorry, can‘t hear you, here under the sheets, crying :sob:

You use track if you can’t cycle onto road lane, the whole point of mapping separately is to have more freedom as to pick if and when a person on a bike can change lanes

Ok, the vote is pretty fifty-fifty with a slight tendency to track :wink:

I think these arguments convince me to rather recommend track:

  • lane suggests that you can change lanes while driving, but protection elements usually prevent this - also for cyclists. Because to change the lane, you usually have to slow down and go through a gap. So - as @Nakaner points out - these protective elements have a similar effect as kerbs.
  • The Wikipedia article shows that these cycleways are rather called cycle tracks in English anyway, which I was not aware of (I only speak English as a foreign language).
  • There are cases where the cycleway is structurally “on the carriageway” or at least on the same surface as motor traffic, but most mappers would call and tag them a track. This makes a strong case against defining "lane for anything that is on the carriageway".
  • Classification as lane or track should not depend on rather subjective characteristics such as “how easy is it to drive over the protection elements with vehicle XY”, because this would not resolve the fuzziness in the categorisation but rather increase it.
  • Mapping Protected Bike Lanes as lane contradicts the existing definitions of lane (=only markings) and track (=there is a physical separation) - the reason why I started this thread.

I would therefore propose a more explicit definition for a cycle track:

A cycle track is separated from the road by curbs, parking lots, grass verges, trees or another physical barrier that prevent an ordinary change to the road lanes while cycling, but is running parallel to and next to the road.

This would still allow the freedom to use lane if, for example, the separation is just provided by e.g. studs.

I think there is a contradiction here. You have to decide if the physical barrier prevents lane change (like a raised kerb, fence, parking bay) or if it is enough if it is only hindering lane change (like a bollard or stud).

By “while cycling” I meant “at full speed”, i.e. in the same manner as one normally changes lanes on a free road e.g. to turn. But it is difficult for me to express this precisely in English…

Whether a change onto the road is physically possible or not, i.e. at very slow speed, is a difficult criterion in my opinion, because it is once again depending on too many (also subjective) factors. (For example, one can also move a large heavy cargo bike across a green strip between a track and the road…).

Despite what Wikipedia says, I’m a native speaker and I’ve never heard these called cycle tracks in the area were I live. Protected bike lane is the term I hear used. Of course that’s because English has many regional variations around the world. The terms cycleway, cycle track, and cycle lane really aren’t used at all in my region and I only know them through OSM. Bike path, bikeway, and bike lane are the terms I hear used.

1 Like

Warning mix of explaination and rant incoming…

Part 1 of 3: It’s not a ‘track’, before you judge see Part 2.

I think “Track” is a bit of a case of England/Euro English and American English being separated by a common language.

Behind track as like railroad track the second most common definition for track is:
“a piece of ground with a special surface for people, cars, etc. to have races or to drive on,” followed by, “(North American English) the sport of running on a track”

The next highest use of the work is… “a rough path or road, usually one that has not been built but that has been made by people walking there” which it’s most common use being (British English).

Yes, I realize that the preferences for OSM are British English but this is a matter where it provides more confusion than clarity.

This is not helped by the fact that when you search for track in OSM iD editor the icon is that of a cyclist with a “racing” helmet and is immediately followed by a person running on a racing track and a horse racing on a racing track.

Then on top of that leisure=track talks about track as something for “running, cycling and other non-motorized racing”

Part 2 of 3: It’s not a lane
No I’m not trying to obtusely contradict myself again I’m just going off what OSM says.

If you look at Key:lanes

The lanes=* key should be used to specify the total number of lanes of a road.

And the following lanes should be excluded :

  • Bicycle lanes. Use the tag cycleway=lane for those.

Note: It also excludes lanes used for parking (no word on if parking is temporary if that counts or not).

Oh but there is an exception…

bicycle lanes only counts for tag lanes=* if the cycleway has a physical separation and it is therefore mapped as a separated OSM way

Bonus: The highway=cycle on the lanes page is immediately followed by a race track with leisure=track + lanes=8 just to cut down on the confusion on Part 1,right?

Part 3 of 3: I think the picture above breaks OSM

Option 1: How this breaks cycleway=lane
How cycleway=lane works today is three ways:

  1. cycleway=lane - means this a lane along each edge of the road
  2. cycleway:right=lane - means there is a cycle lane on the right edge of the road
  3. cycleway:left=lane - means there is a cycle lane on the left edge of the road

The theme here is that cycleway=lane presently must always be on the edge of a road. So a cycle lane between the through lane and right turn lane (if you drive on the right side of the road opposite if your don’t) isn’t something cycleway=lane as presently implemented to handle. Which is ironic because this is the safest way to do a bicycle lane bordering a turn lane and by in large the best practice.

Bonus issue: Because cycleway=lane is not a counted lane in lanes if a bicycle lane crosses other lanes you can’t use Relation:connectivity either to connect it as the lane to it’s equal at an intersection. Which can be problematic if you have an intersection like this: Mapillary

Which Then a type=connectivity from Way 1065994301 to Way 1065994305 would read connectivity=1:2|2:3|3:1

Option 2: How this breaks the methodology of a matrix-grid-columns whatever you want to call it tagging

In the quoted example provided and based on the example given in Lanes - OpenStreetMap Wiki (Specifically the Crossing with a designated lane for bicycles section) the tagging would look like this…


This looks right but if there are lanes=3 and I’m a renderer or router would I expect 5 lanes? Most probably not which means that this method likewise can’t as presently support a bike lane between a through and a turn lane much less anything more exotic.

Heck you can’t even tell how many cycle lanes there are either because cycleway:right=lane + oneway=no implies one lane not two so you can’t even infer there might by more than 3+1 much less five.

The best solutions would be for OSM to treat all traffic as traffic and all lanes as lanes, this sound intuitive but the roots of OSM being vehicle in emphasis first has set in deep. Could you imagine OSM being able to capture everything in the entire right-of-way though? Bike lanes, sidewalk, buffers, parking, both inside the curbs and outside? Like the ability to in iD or JSON to edit with a top and side view of the whole right of way? Wouldn’t that be an absolute joy?

More realistically though… cycleway needs a option for a turn lane crossing the bike lane for the merge zone and likewise an option for the bike lane one it’s been crossed to cover the traditional two phases where bike lanes and turn lane crossover one another.

1 Like

I wanted to expand on the concept of all lanes given equal treatment as lanes and what options this could open up.

Let’s say I want to be able to present a full cross section of a road similar to the one I have here on Streetmix

Today I would have to map each sidewalk x2, because the roads are separated by a median the roadway x2 and I’d have a cycleway on the right and parking on the right but very little way to list which is the furthest right because bike lanes can be on either side of parking.

Now consider all lanes are lanes and we use the proposed separation tagging, as well as the buffer tagging, only the separation/buffer has one less column as lanes because it’s always describing the space between lanes, otherwise we’ve had to list a lot of things twice.

Using the present tagging design with very little changes the above example would be tagged like:


Then everything here has lane-1 options because they describe the space between lanes:


Note: I totally get that a LOT of this doesn’t exist today this is not a huge effort on my part to “fit in” as much as it it is to get people to imagine and see what OSM has the potential to be and dream a bit. With that in mind, dream with me.

What possibilities would this open up?

  • You not have a single way instead of 4 separate ways allowing for a HUGE amount of consolidation
  • This would allow for iD and OSM to render both:
  1. a top down visualizations of the entire right-of-way, not just the portions between the kerbs
  2. a simple profile view like one that Streetmix has with a much more intuitive drag and drop approach with a few toggles and switches.
  • cyleways no longer have to be locked to road edges and can be arranged anywhere that makes sense
  • if a road or bike lanes borders street parking it’s clearly known as is which side and how much buffer if any is provided
  • sidewalks become part of the right of way
  • the type of separation between things is known
  • things that are only presently open to vehicle lanes can be expanded and reused for cyclelanes and even sidewalks like:
  1. type=connectivity - where you can specify how complicated lanes connect-cross
  2. restriction=* - limiting certain turn types
  3. routing can tell cyclists and motorists both about upcoming interactions that they can’t se are related today
  4. lighting could be more easily added is a visual edit
  • you don’t end up with miles long suffix combinations to handle things like cycleway:right:separation:left:width=*
  • this addresses somethings like the current propose is_sidepath=* or cycleway=separate

Sorry for the long posts but the is a complicated issue that I’m starting to come to grasp with as a mapper and I’m trying to throw out these ideas and see what people think.

I’m not actually sure if you are joking or not, but this is unmaintainable. It’s close to impossible for humans to read, let alone spot errors in this tagging. As much as I like the way of thinking, this would require some serious support from all editors to be even remotely usable. Plus, it would cut the streets into tiny little fragments.
I don’t think that this is the future, but “Well, you know, that’s just like uh, my opinion, man”
I hope someone got the reference. I’ll see myself out :door: