routing problem - motorway vs motorway_link
  • This has been mentioned earlier but I can't find it back.
    Tested with MNF 1.5.4

    Coordinates 
    51.41004/6.12479 to 51.35606/6.14548.
    It is the A73 from North to South. At that point there is quite a long parallel traject being a motorway link. Max speed is 130 for both the motorway and the motorway link. MNF leads you over the  motorway_link instead of the motorway, most probably because the motorway_link is the "inner bend" and therefore probably shorter, but this is not correct of course. (but the OSRM map-project site does the same).

    image

    I did have this error before. Most often in Germany where theser kind of motorway constructions are more common. It is obvious that MNF treats motorways and motorways_equally.

    I also know that you are currently working on the routing, but how does that deal with these kind of <road> versus <road>_link
  • 25 Comments sorted by
  • "with default car settings it is 1s faster on the link road" 

    Yeah, that's because the link is the "inner bend" and the motorway is the "outer bend". That's exactly my point: MNF makes no difference in a <road>_link versus a <road>. 
    MNF should give a slightly lower priority to a link.
  • The OSM data has not changed. The map creation has changed. MDX is now using some algorithm to simply set the speed of all slip roads/ramps/exits to 50% of the main road. He did that for a number of important countries as an experiment. I can't remember Austria was one of them but I suppose it is.
    I'm eagerly waiting for the availability of the Netherlands.
  • I like this idea. It may not be a complete fix but it will help greatly

    I hope the US will be done.
  • In most cases, when MF uses link roads, that there was a main road with speed limit and link road has no indicated speed limit. So the link seems to be faster than the main road.

    http://forum.mapfactor.com/discussion/410#Item_11
  • Thanks for pointing me to that post.
    However, in this case it is just the other way round: both the link and the motorway have a maxspeed of 130 for the firsrt half of the traject. And 100 km for the second part of the section.
  • with default car settings it is 1s faster on the link road

  • Maybe this is not the link?
  • I assume you mean the error in the 2014 maps that mdx is mentioning.
    I checked for a new Dutch map and the 44-201501140 was available for download.
    So I downloaded that one and checked again: It doesn't make a difference. The routing is still over the motorway_link.

    I mentioned: "MNF should give a slightly lower priority to a link."
    Optimally this should be configurable, but it can even be hardcoded.
    When calculating a route it needs to evaluate all road types anyway. Currently there are preferences and speeds for every road type, meaning that the A* pathfinder algorithm (I can't imagine MNF is using another algorithm) is using penalties anyway (road preferences and road speeds), next to the "heuristic coefficient" which has influence on speed of calculation, memory and accuracy of the route.

    MNF could easily do something like "preference = preference * 0.9" or something alike for a link.
    More navigation apps use the exact same approach and calculate it correctly. It probably means a slight overhead in the calculation, but it would give more accurate routing.

    OSRM is using a completely different approach and algorithm for its routing.
  • Maybe the same kind of problem: Not a motorway, but "primary" vs. "primary_link":

    Driving on B178 from west (47.48249 12.122607) to east (47.482744 12.129731) should follow the B178 (primary) below the roundabout but uses the primary_links and the roundabout. Although the links don't have speedlimits I can hardly believe that this way is faster.

    That's how it should be:
    http://map.project-osrm.org/?hl=de&loc=47.482460,12.122678&loc=47.482708,12.128820&z=17&center=47.483112,12.124014&alt=0&df=0&re=0&ly=763558683

    OSM:    http://www.openstreetmap.org/way/12968187#map=17/47.48289/12.12629&layers=N

  • Yes, MNF does that for all road types.
    That's also why I mentioned in one of the previous posts:
    "MNF makes no difference in a <road>_link versus a <road>"
  • There are no maxspeed tag on rounabout and links.

    I had almost different issue with links, with too low priority and routing throught the 'center' of crossroads in right turn.
  • @mdx: This discussion about <road>_links versus <road>s: Is MNF investigating this and working on this?

    And what about my suggestion 'MNF could easily do something like "preference = preference * 0.9" or something alike for a link.' which I mentioned in this post. Is that feasible?

    Edit: The fact that I'm pushing this is that it generates a relatively unsafe situation. MNF tells you take the exit while you can clearly see on the road signs that you have to go straight on. What's even more confusing is that the MNF road signs on the screen, with other destinations belonging to that exit, also tell you to take the exit obviously pointing you to a direction you certainly don't want to go (even though it leads you back on the road later on).
    Confusing situations are unsafe situations.
  • See my comment & link above. I noticed in my settings that motorway preference was set to 81% I changed it to 65% & since then it hasn't tried to take me onto the link road rather than the motorway. Not sure if there is a connection but it may be worth trying.
  • I always drive with motorways at 60%. I even tried, like you are suggesting, switching it back to 50% and up to 75% even though I actually know this is not going to work: For MNF a motorway_link is identical to a motorway. They have the same preference. Changing the preference is changing it for the both: no solution.

    I would be very surprised if this works for you. 
  • As also freshly referred to in the other post: This is still not solved!

    This could be solved in a few ways as it is done in other nav apps as well (3 solutions proposed):

    1. priorities: slip roads / link roads have lower priority then main roads. If the current map setup does not allow this, it could be hardcoded (until the map structure is adapted). The road tags need to be read anyway (directions, speed, lane assistant, etc.), so if it is a link road set the priority to 70% - 90% of the main road.
    2. Speed penalty: If available use the road speed of the link road / slip road. If no speed given for the link road use 70% - 90% of the speed of the main road for the link road. If speeds are equal still use 70% - 90% of the speed of the main road.
    3. Time Penalties: When calculating a route always recalculate the linkroad time to "link road travel time + 20 seconds) , or something like that. 


    In one of the other (repeating) topics about traffic time calculations with regards to traffic lights, traffic_calming and so on (here and here), I also mentioned  the time penalties option 3 in the second thread.

    Priorities should however be the correct routing over the main road and not the slip road/link road.
  • I checked it once again and everything seems to be fine.
    With my preferences road/routing goes through main lanes (motorway).
    I still think that this separate lane of motorway is not link - it is just comfort additional lane for thous, who want to use link - turn left/right. If drive this road/lane may be more dangerous than 'main' lanes, there should be speed restriction - question for owner, why it is not.

    How to solve it - just add one second for each crossroad - if there is joined node, one sec. more.

    Additional, for eco route, each left turn tagged with 30 km/h and each right turn 40 km/h.

    The next case - roundabout - there are still no speed restrictions on roundabout and links - EDIT - Potlatch - klik on speed sign and save.
  • Can anyone of the support team or one of the Devs say something about this?
    Has this now been solved? And if so: In which version?
  • No, it is not solved, at least not my example from above (Jan 25) and v. 1.6.14:

    <set>
    <name>zTest B178_1</name>
    <departure><name>B178 W</name><lat>170936856</lat><lon>43641641</lon></departure>
    <destination><name>B178 E</name><lat>170937749</lat><lon>43663752</lon></destination>
    </set>

  • Looks interesting: The result of my example two posts above has changed with the latest map version of Austria (12.06.2015): The routing is now correct. As far as I can see there has not been any change in OSM-data. See screenshots below may vs. june: (I am not able to insert screenshots)
  • I found it back. It's this post: http://forum.mapfactor.com/discussion/comment/10299#Comment_10299
    And yes: it seems for all countries in the June release
  • I needed some good news.

    Thank You

Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

In this Discussion