Cycleway routing problem: long detours to avoid altitude gain, no distinction of shared vs dedicated #5123

New issue

Open

opened 2023-05-11 07:46:17 +00:00 by ellieayla · 0 comments

ellieayla commented

2023-05-11 07:46:17 +00:00

(Migrated from github.com)

Describe the issue
Cycle directions in OrganicMaps, Graphhopper, OpenStreetMaps.org (with Graphhopper directions), and in OsmAnd avoid the north end of a nice cycle track.

Steps to reproduce

Open OrganicMaps
Start navigation from 43.33461, -79.81088
End navigation at 43.315268, -79.801664
Select cycle directions. Observe mostly following Brant Street. Screenshot 1
Select foot directions. Observe following a fancy multiuse path tagged bicycle=designated. Screenshot 2.

Expected behavior
Maple Trail (way 842437412) is vastly preferable above the parallel streets. It is tagged bicycle=designated. But doesn't appear to be prioritized in the router.

Screenshots

Investigation

I figure there's something weird with way tagging, so I tried to narrow it down by finding a smaller sample on OpenStreetMap.org, and I get an even weirder case.

Go to https://www.openstreetmap.org/directions
Set navigation start point to 43.33401, -79.81041
Set navigation end point to 43.3271, -79.8089
Set navigation use to Bicycle (Graphhopper)
Observe route follows Ghent Avenue, additional screenshot 1
Move navigation endpoint slightly west, to 43.3337, -79.81067
Observe route follows Maple Trail, additional screenshot 2
Repeat in OsmAnd Maps, observe very similar, additional screenshot 3

Additional screenshots

System information:

Operating system and its version: iOS 16.3.1
Organic Maps version: ⁦2023.05.08-1⁩ (with OSM data May 3, 2023)
OsmAnd version: 4.4.7
Device Model: iPhone11

Additional context
Given the weirdness across various apps, I don't think this is an OrganicMaps issue. It's probably a OSM data issue. Or something common to the routers. But I cannot determine why this trail doesn't suck in cycle traffic. Maybe turn restrictions, maybe one of the ways is tagged weird, maybe the bollards are discouraging bicycles.

I edited this way recently, so it's entirely possible (likely even) that I introduced any such data defect.

https://fosstodon.org/@organicmaps/110348752285013846 suggested opening an issue here anyway, presumably as a starting place for figuring out what I've done wrong.

I humbly seek your guidance.

**Describe the issue** Cycle directions in OrganicMaps, Graphhopper, OpenStreetMaps.org (with Graphhopper directions), and in OsmAnd avoid the north end of a nice cycle track. **Steps to reproduce** 1. Open OrganicMaps 2. Start navigation from 43.33461, -79.81088 3. End navigation at 43.315268, -79.801664 4. Select cycle directions. Observe mostly following Brant Street. Screenshot 1 5. Select foot directions. Observe following a fancy multiuse path tagged bicycle=designated. Screenshot 2. **Expected behavior** Maple Trail (way 842437412) is vastly preferable above the parallel streets. It is tagged bicycle=designated. But doesn't appear to be prioritized in the router. **Screenshots** ![2A58735D-269B-4FE4-85D6-55D71BF16D28](https://github.com/organicmaps/organicmaps/assets/1447600/c029319c-8e1c-4453-8014-0bfaf0578f1b) ![4F4BC98E-DF59-4B83-8E0C-3F70DC7D7D7C](https://github.com/organicmaps/organicmaps/assets/1447600/e6ddca0c-2127-41c0-9d91-3d12ed3becaa) **Investigation** I figure there's something weird with way tagging, so I tried to narrow it down by finding a smaller sample on OpenStreetMap.org, and I get an even weirder case. 1. Go to https://www.openstreetmap.org/directions 2. Set navigation start point to [43.33401, -79.81041](https://www.openstreetmap.org/#map=19/43.33401/-79.81041) 3. Set navigation end point to [43.3271, -79.8089](https://www.openstreetmap.org/#map=16/43.3271/-79.8089) 4. Set navigation use to _Bicycle (Graphhopper)_ 5. Observe route follows Ghent Avenue, additional screenshot 1 6. Move navigation endpoint slightly west, to [43.3337, -79.81067](https://www.openstreetmap.org/#map=18/43.3337/-79.81067) 7. Observe route follows Maple Trail, additional screenshot 2 8. Repeat in OsmAnd Maps, observe very similar, additional screenshot 3 **Additional screenshots** ![F402E781-1EBE-4736-AF6A-457EB04DB8B1](https://github.com/organicmaps/organicmaps/assets/1447600/327f8e80-093d-4004-b05c-119d39f140e2) ![8BC7D3F7-1717-49B9-A16A-9237ED42A252](https://github.com/organicmaps/organicmaps/assets/1447600/dc5522a5-47b6-41d9-b053-4b6776697e1b) ![83F27F34-C31A-40FB-8843-14D33E149DFD](https://github.com/organicmaps/organicmaps/assets/1447600/16fa197e-7eaa-4ca6-8a7b-7519af078774) **System information:** - Operating system and its version: iOS 16.3.1 - Organic Maps version: ⁦2023.05.08-1⁩ (with OSM data May 3, 2023) - OsmAnd version: 4.4.7 - Device Model: iPhone11 **Additional context** Given the weirdness across various apps, I don't think this is an OrganicMaps issue. It's probably a OSM data issue. Or something common to the routers. But I cannot determine why this trail doesn't suck in cycle traffic. Maybe turn restrictions, maybe one of the ways is tagged weird, maybe the bollards are discouraging bicycles. I edited this way recently, so it's entirely possible (likely even) that I introduced any such data defect. https://fosstodon.org/@organicmaps/110348752285013846 suggested opening an issue here anyway, presumably as a starting place for figuring out what I've done wrong. I humbly seek your guidance.

👍 1

This repo is archived. You cannot comment on issues.