Utilizing route=bicycle relations in long-distance routing
As we all know, bicycle routing is hard. Unlike motor vehicle traffic, where there is a clear hiearchy of roads, the bicycle road network consists of all sorts of roads. In my area, there are no clear 'bicycle trunk roads' or 'bicycle motorways' that could be used as the backbone of a longer journey. I guess that the situation could be better in more bicycle-friendly countries, such as Denmark or the Netherlands. I was thinking that maybe we could tweak the bicycle routing by increasing the weight or speed of ways that belong to a route=bicycle relation, or by defining 'wormholes' between some points of the relation. I guess that this is not doable in the default style, which aims to be 'neutral' for all modes of transport, but I guess it could make a huge difference in a bicycle-oriented map. FWIW, I just defined a 25km route relation that could be useful for testing: http://www.openstreetmap.org/browse/relation/3252216 IMO, this is much better than the NCN 4 in this area. Has anyone experimented with this kind of tweaks to routing? Would this bypass the limitation that the Garmin Edge 705 seems to ignore cycleways for the middle section of a route? Marko
Hi Marko, Yes I already do those tweaks in my Openfietsmap. You can test your route when it is rendered sooner or later on Lamertus' server http://garmin.openstreetmap.nl Every way that is part of a relation with route=bicycle gets a higher road class/speed, so when you choose to set the routing to faster route instead of shortest way, it takes a more scenic bike route. If you use "avoid toll roads" the routing is forced to use those route=bicycle relations.
Has anyone experimented with this kind of tweaks to routing? Would this bypass the limitation that the Garmin Edge 705 seems to ignore cycleways for the middle section of a route?
I dont know the Edge 705, what I do know is that the new Edge Touring now doesn't have a car/pedestrians settings anymore (only mtb/touring bicycle and bicycle mode), and also no options to avoid toll roads. So when users want to use my map this really sucks :( BTW The Edge Touring is a very interesting device because it contains an official Garmin EU cycle map based on OSM!
Hi Marko,
FWIW, I just defined a 25km route relation that could be useful for testing: http://www.openstreetmap.org/browse/relation/3252216 IMO, this is much better than the NCN 4 in this area.
I have found in wiki that route=bicycle defines "named or numbered or otherwise signed routes". Wouldn't people expect some markings on roads, if they find a route on a map?
Has anyone experimented with this kind of tweaks to routing?
I have tried to make different routing overlays for different activities. It is not easy, since Garmin defines many new activities, which aren't supported by mkgmap. I think currently only car mode allows for tuning of routing by changing route class and speed. This way you can get more suitable route but usually with wrong ETA. -- Best regards, Andrzej
On Tue, Oct 08, 2013 at 11:47:15AM +0200, Andrzej Popowski wrote:
Hi Marko,
FWIW, I just defined a 25km route relation that could be useful for testing: http://www.openstreetmap.org/browse/relation/3252216 IMO, this is much better than the NCN 4 in this area.
I have found in wiki that route=bicycle defines "named or numbered or otherwise signed routes". Wouldn't people expect some markings on roads, if they find a route on a map?
Yes, and that is one reason why I posted about this to the "Users: Finland" forum at <http://forum.openstreetmap.org/viewforum.php?id=15>. I do not think that this is as bad a sin as "mapping for a renderer", but it does feel a bit wrong. On the positive side, maybe this route could some day become part of the official route.
I have tried to make different routing overlays for different activities. It is not easy, since Garmin defines many new activities, which aren't supported by mkgmap.
Oh, I guess that this is the "activity routing", only available with newer devices? I more or less abandoned the Garmin after switching to a rugged Android phone (SonyEricsson Xperia Active) almost two years ago, after the USB data transfers on my Edge 705 stopped working. OsmAnd is good enough for displaying map and searching for POIs. Its bicycle routing sucks too, but at least it is updating the map while recalculating. FWIW, there is a stand-alone offline routing program for Android that can be used with OsmAnd. Unfortunately it appears to be closed-source and on data that has been converted with an undisclosed tool: <http://brensche.de/brouter/>. It claims to support multiple routing profiles (prefer cycle routes, prefer flat terrain, etc.). I did not experiment with it yet, because I prefer openness. Best regards, Marko
Am 08.10.2013 13:07, schrieb Marko Mäkelä:
Yes, and that is one reason why I posted about this to the "Users: Finland" forum at <http://forum.openstreetmap.org/viewforum.php?id=15>. I do not think that this is as bad a sin as "mapping for a renderer", but it does feel a bit wrong. On the positive side, maybe this route could some day become part of the official route.
Hi Marko, if this route does not belong to an cycle network please don't use network=*cn or *cn=yes, which are defined for bicycle routes which belong to such a network. In general I think there are enough bicycle routes which you can use for testing ;) Henning
On Tue, Oct 08, 2013 at 06:54:57PM +0200, Henning Scholland wrote:
if this route does not belong to an cycle network please don't use network=*cn or *cn=yes, which are defined for bicycle routes which belong to such a network.
I did not use any of network=*, *cn=* or ref=* and do not plan to.
In general I think there are enough bicycle routes which you can use for testing ;)
Maybe. This one I could test in real life, because I often ride to places that are close to this route. Another problem related to cycle routing is the relationship between cycleways and the nearby roads. Sometimes cycleways run in parallel to a motor-vehicle-only road, but in bigger intersections they can get further apart. Around here, the cycleways are usually unnamed. Only in some rare cases there is a name for a cycleway/footway. One problem with these unnamed cycleways that run in parallel to a named main road is that the name will be missing in navigator prompts. It would be nicer to be prompted to turn to "Main Street" or "Main Street (cycleway)" instead of an unnamed way. But, it is arguably wrong to add the name tag to the cycleway, if it is not a lane of Main Street. Also, I guess that for long-distance routing outside urban areas, it would be good if the routing engine followed a highway=trunk to get the "rough picture" and then added "micro-routing" for the nearby cycleway based on the relationship of these two ways. For the purpose of bicycle routing, the cycleway is "equivalent to" the motor vehicle road. But, how to tag this in such a way that mkgmap and routing engines can easily make use of this? The closest relation types that I can find on the wiki are 'site' and 'street', both only in proposed status, and neither is for exactly this use case. Best regards, Marko
participants (4)
-
Andrzej Popowski -
Henning Scholland -
Marko Mäkelä -
Minko