New tendency to route through hiking paths

[Jean-Alexis Montignies]

Hi,

On brouter-web, I noticed I've been directed into impractical routes (dirty hiking paths with 15%+ slope), very hard, even when pushing a non loaded bicycle.

Checking on the map, it looks like those path are indeed marked as hiking paths.

I think there's something wrong here.

Did the algorithm changed the way it handles that in the recent months?

I can post some examples if needed.

Thanks,

Jean-Alexis

[Poutnik Fornntp]

The best is always providing in any such a post the current  Brouter web URL link, which includes start, destination, viapoints, no-go zones and used built-in profile. 

Optionally, attach or refer to a used custom routing profile, if applies.

There is always need to check, if the issue is caused by incorrect mapping somewhere(frequently is), by the used profile, or by some Brouter internal bugs.

Dne 26. března 2021 22:29:52 Jean-Alexis Montignies <ja...@montignies.info> napsal:

--
You received this message because you are subscribed to the Google Groups "OSM Android bikerouting" group.
To unsubscribe from this group and stop receiving emails from it, send an email to osm-android-biker...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/osm-android-bikerouting/e6ea24f5-023d-4df1-841e-e39073ea7593n%40googlegroups.com.

[Jean-Alexis Montignies]

Right, that was just a general feeling post :). I stumble on impossible routes (I was using the standard cyclotourist profile on the web application)

I checked in OSM database and it seems the bicycle flag is set to YES, although it's a steep path. In one case I'm sure it's for MTB, the other one not sure.

So this is problematic: how can we distinguish between routes ok for MTB but not a standard bicycle?

A short flat path is still OK for a standard bicycle. 

Here are some examples:

http://brouter.de/brouter-web/#map=16/45.7115/6.9297/OpenTopoMap&lonlats=7.317928,45.74263;4.797577,45.296305

Km 44:it’s a MTB track 10-15%., no way a cyclotourirst could ride there.

So there is a conflict in mapping (or should the mtb-uphill difficulty limited)?

http://brouter.de/brouter-web/#map=15/45.9016/7.0773/OpenTopoMap&lonlats=7.080352,46.107719;6.986618,45.772799

Same problem here km35. Maybe a mapping problem (bike flag is yes). But if bicycle are allowed, the cyclotourist profile should not allow to use these routes.

Jean-Alexis

[Poutnik Fornntp]

I think this behaviour of the standard trekking profile was since beginning. It does not discriminate too steep climbing, as it does not penalize climbing at all.

It penalises downhill rides >1.5%. The rationale behind is that only in this case you waste energy,  via excessive air drag ať high speed or via breaking. Climbing may be uncomfortable, but you accumulate energy, reusable for mild descends. ( This is true for mild climbing, but becomes less true for steep climbing )

While this is reasonable for many cases, it may fail in extreme or unusual scenarios like riding near Mont Blanc. If there are 2 alternatives, one shorter but steeper, it may be chosen, depending on distance ratio versus cost ratio. 

E.g , just illustratively,  if there are 2 alternatives to reach a saddle,   the path 5 km long 16% steep, with path costfactor 5 ( 1 km like 5 km of an ideal way ), or hairpin-like primary mainroad 10 km long with costfactor 2 and steepness 8%, the path would be chosen and not the road. As 10*2 < 5*5.

Cost evaluation can be observed on the web under "Excel sheet" icon.

One should not expect any profile to work always ideally or along our expectation. One should watch and make manual adjustments of the route if one feels one knows better.

Main workaround and suggestions:

1) manual route adjustments by route dragging

2) Applying additional climbing penalisation, discriminating steeper hills by tweaking the profile on the web. Favourite one is e.g

assign uphillcutoff 3.0

assign uphillcost 70

( do not add, modify existing one )

3) Trying custom profiles, uploading them in the profile upload section, that use different altitude and cost evaluation, 

see https://github.com/poutnikl/Brouter-profiles/wiki

Dne 27. března 2021 21:58:45 Jean-Alexis Montignies <ja...@montignies.info> napsal:

To view this discussion on the web visit https://groups.google.com/d/msgid/osm-android-bikerouting/fb618609-09ec-4a1a-8811-2aefa4a427d6n%40googlegroups.com.

[Volker Schmidt]

I think this is a mapping problem in OSM.

The "culprit" seems to be OpenCycleMap rendering in combination with mappers who want to see MTB suitable paths on the OCM.

OCM renders highway=path with bicycle=yes as a cycle track (dashed blue line).

Take a look at the "cycleways" in  this map area in the Colli Euganei and the corresponding Mapillary Mapillary images. (example)

These are tough hiking paths, often badly maintained, very popular with MTB fans.

I think the idea of OpenCycleMap to show highway=path and bicycle=yes as a cycleway is wrong, at least in this context, but there may be other parts of the worls wher this si the standard way of tagging unpaved cycleways - I don't know.

Also the mapping may be wrong. This path, partially on rocks, is mapped MTB_scale=5, and the Italian-langage description says, correctly, that there are points too narrow for MTB.

The problem is that bicycle=yes is not wrong, because, to my knowledge, these paths are not closed to bicycles in the sense of legal access.

To view this discussion on the web visit https://groups.google.com/d/msgid/osm-android-bikerouting/fb618609-09ec-4a1a-8811-2aefa4a427d6n%40googlegroups.com.

[Poutnik Fornntp]

P.S.: Saying below, it adds to the mapping issues Volker mentions. I did not have my laptop handy to evaluate it properly, posting from my phone.

Dne 28. března 2021 8:30:32 Poutnik Fornntp <poutni...@gmail.com> napsal:

[Jean-Alexis Montignies]

I feel the same way as Volker, it’s probable that the decrease in quality of the BRouter itineraries are caused by those MTB tracks.

Of course we can not ban mountain bikers to mark path as authorised to bicycles.

https://wiki.openstreetmap.org/wiki/Mountain_biking

Maybe path with mtb:scale >1 and mtb:scale-uphill >0 should be excluded for trekking / commuting. 

I can always go through all the length of calculated routes to exclude those paths, but it's then really tedious. I used to mainly trust brouter and not checking in details the routes.

As for the uphill cost, it makes sense as it is calculated now.

Except for the next problem: 1km at 20% is much more harder than 2km at 10% (I don't know for you but at least 5 times).

How is calculated the cost of uphill/downhill? Is it dependant on the slope or the elevation change? Is it proportional or just the length of the ascent, descent? (Well I could take a look at the source)
I'm sorry I'm new on this forum, this may have already been discussed before.

Jean-Alexis

[Poutnik Fornntp]

Not that built in profiles do not evaluate MTB scale tags.

My profiles from https://github.com/poutnikl/Brouter-profiles

evaluate MTB scale  by progressively   growing penalisation, but ignore MTB uphill tags.

Profiles are not fool proof  If they were, they would be very complex, close to AI systems.

I have previously said by other words the energy gaining is not equivalent for steep slopes.

for computing of elevation penalisation, see

https://github.com/poutnikl/Brouter-profiles/wiki/Glossary.

Dne 28. března 2021 14:19:49 Jean-Alexis Montignies <ja...@montignies.info> napsal:

.......

Maybe path with mtb:scale >1 and mtb:scale-uphill >0 should be excluded for trekking / commuting. 

..... I used to mainly trust brouter and not checking in details the routes.

As for the uphill cost, it makes sense as it is calculated now.

Except for the next problem: 1km at 20% is much more harder than 2km at 10% (I don't know for you but at least 5 times).

....