Unrealistic represenation of small residential roads

[Robinlovelace]

Here's an example: https://www.openstreetmap.org/way/476939396

In practice it is a small tarmac road with concrete bricks as pavements. There are for sure no junction signs or anything here, they are tiny roads, it's an internal road. Reported by Varad. Links with #151 and other issues around junction signposts. Will aim to add screenshots.

[Robinlovelace]

Having looked at this I think this links to a wider issue that Dustin has raised before: many roads on which you can travel both directions have sufficient room for 2 way traffic. Even more roads lack a central line marking. Upstream question I guess is how to mark such roads on OSM, using the 'width' tag?

[Robinlovelace]

[Robinlovelace]

Another example: https://www.openstreetmap.org/way/6273624#map=18/53.81899/-1.53868

[Robinlovelace]

Here's how it's represented in A/B Street (not subtle grey shading for living street), really they could all in this 'LTN' area be defined as living streets (although definitions vary) but only Sholebroke Mount marked as such on OSM (by me):

[Robinlovelace]

Roughly the same area in A/B Street. Liking the colouring (although would have a mild preference of grey green for the living street, think this has been discussed but cannot recall the reasoning for grey pink over grey green).

[Robinlovelace]

This specific issue becomes clear when you zoom in:

No way that there are Stop signs all over the place, not sufficient room for 2 cars side-by-side, there is car storage on both sides, and there is in reality no centreline. This is a common road configuration in many dense urban areas built before garages for car storage became common in the 1960s or so. I think the first question is

• How to tag this on OSM?

Then we can think about visualisation. My first thought, however, is that residential roads should not have central lines or stop signs by default except perhaps in US (and Canada?).

[Robinlovelace]

Also note: Back Sholebroke Mount is tiny and is actually a bit of a kids play area. It lacks a pavement and is really a thin access only road often blocked, if one person parks on that street no cars can pass creating temp modal filters much of the time.

[dabreegster]

Thanks for reporting! There's a few different problems/ideas here:

1) The inferred width of the road is obviously wrong, because it smushes into buildings. In #870 I'm taking first steps towards detecting similar situations and just shrinking width. There's a bigger question of how to represent / infer more reasonable road width, one that I'm exploring on March 12. So, there are some immediate steps I know how to take here -- just need time to work on them.

2) Road center-line markings are just for rendering right now. It sounds like there's interest in changing their style based on region -- at the very least, yellow center lines aren't used in the UK for sure. I'm interested in doing this, but also slightly hesitant -- A/B Street tries to have a schematic rendering. The center-line tells you explicitly that there's two-way traffic. Even if that's not physically the marking on the ground, it might be useful to communicate this.

3) Too many stop signs. (I've seen like 3 stop signs physically in the UK so far!) Maybe we need yield markings instead. Again, the purpose of rendering is partly to communicate the rules that agents in the traffic simulation will (theoretically) follow. So we could consider different rendering style for the UK and/or improve heuristics for stop sign placement, or even use data from OSM. #795, #781 are related.

[Robinlovelace]

Sounds good and agree that there different things going on. Genuine unanswered question though:

How to tag roads on which traffic can go in either direction but which are so thin as to be able to only accommodate 1 way travel (2 cars cannot fit side-by-side)?

That is not just a cosmetic issue because it would affect the traffic flow, slowing it down. Idea on that: halve car speeds on such roads and allow them to 'ghost' past each other to avoid representing complex pull-in/waiting maneuvers. There has been some research into potential benefits of removing road centrelines (Shackel and Parkin 2014). That suggests to me it could be worth visulising roads that don't have them differently (acknowledging the fact that from OSM tags alone we rarely know, but could make perhaps make educated guesses, e.g. if the road width is less than say 6m).

[dabreegster]

Quick-and-dirty hack: https://wiki.openstreetmap.org/wiki/Key:narrow

If there was a sane editor workflow for mapping and satellite coverage is good enough, https://wiki.openstreetmap.org/wiki/Key:width

[hungerburg]

In JOSM, there is a rendering style plugin, it is called https://josm.openstreetmap.de/wiki/Styles/Lane_and_Road_Attributes, with adequate aerials, you can do realistic measurements of some highway types.