Coincidentally, we've been having a similar discussion in an NO context.
Some combined foot/cycleways physically end at a bus stop or similar,
but the mapped foot/cycleway continues across the kerb to join a
parallel carriageway. Example 1
<https://www.openstreetmap.org/way/121911706>, Example 2
<https://www.openstreetmap.org/way/1422918378>.
The kerbs make further travel impractical for wheeled users, and
impossible for some, so I invited the community to an informal
discussion on the benefits and problems associated with these virtual ways.
My position is that this mapping puts some travellers at risk of not
being physically able to follow the routing instructions, and leaves
them stranded either side of the kerb, whereas the opposing position
seems to be that not mapping in this way means some travellers will not
be given the optimal route.
People are free to see the opportunity to optimize and make the
transition themselves, but the same is not true for the opposite -- to
see that you are being routed into a dead end and left stranded.
I have to point out that we're not mapping roads in these examples, and
there's no acommodation for people to travel along that virtual line
(like a lowered kerb that road builders frequently use), so our scenario
is not how to map a road that has a barrier, but instead it's a question
of mapping a road that doesn't exist -- in spite of there being a barrier.
The argument in favour is mapping a way across the kerb is that some
people are able to pass it. I believe the same argument applies to
mapping ways across two kerbs (like a divider), or across a patch of
grass, a drainage ditch, etc. How would we determine where to draw the
line between a barrier that's passable and one that isn't? I can easily
lift my road bike over a 50 cm tall concrete barrier, but I cannot
traverse a kerb using the cargo bike.
The issue of being routed on a way that's not passable can be reduced if
mappers tag the kerb on a node, and routers respect this and give users
a clear option to disallow crossing kerbs -- or simply decide for them
based on profiles.
To me, this means mapping something that doesn't exist, for the benefit
of some users, and the disadvantage to other users can be mitigated in
theory, but this will not likely happen in reality.
I'd be more comfortable with the idea that routers decide that
travellers can traverse the kerb on a pavement than the idea that we
create ways that don't exist to trick routers into doing something the
router wouldn't normally do.
Jens
On 8/17/2025 8:13 PM, Michael Tsang wrote:
Hi all,
I am having a problem at a place where my router is overestimating the
cycle time required for my journey, leading to bad results by default.
The problem is here: https://www.openstreetmap.org/node/4560492277
When I alight from a train with my bike and want to cycle north-west
after exiting the station at the specified node, in the real world I
would push my bike through the kerb onto the carriageway and start
riding. However, the mapping doesn't indicate that I can do so, so the
router think that I would need to either push the bike along the
pavement to the next junction, or use the signalised crossing to get
to the carriageway. This has resulted in it overestimating my bike
time by about 2 minutes.
What is the correct mapping here to indicate that I can push my bike
to the carriageway and mount it?
Thanks,
Michael
_______________________________________________
Tagging mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/tagging
_______________________________________________
Tagging mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/tagging
