Hi, I have seen some strange routing issues with the way below. If I am driving south, my GPS wants me to make a U-turn where the divided highway ends. It seems like the ways are not connected but I checked and I can't find anything wrong in the OSM data. http://www.openstreetmap.org/browse/way/37649065 Any ideas? Francisco
Hi The fault is in osm-data. There is no turn restriction tagged. I've added such an restriction. Henning
On 2:59 PM, Henning Scholland wrote:
Hi The fault is in osm-data. There is no turn restriction tagged. I've added such an restriction. But why do I need the turn restriction here? I haven't seen this type of routing problem in other roads that have a similar layout where the split/divided ways merge into one. I thought there would be some discontinuity somewhere but I didn't see if the OSM data to explain why the routing was broken.
Francisco
On Sat, Jan 07, 2012 at 09:13:25AM -0500, Francisco Moraes wrote:
But why do I need the turn restriction here? I haven't seen this type of routing problem in other roads that have a similar layout where the split/divided ways merge into one. I thought there would be some discontinuity somewhere but I didn't see if the OSM data to explain why the routing was broken.
Right, that's why I asked where you were heading to and what the routing suggestion looked like. Was the Garmin unit configured to auto-recalculate, and was there some bad coverage (tunnels) that caused it to "jump" somewhere else and back? Or could there be some error in the map data elsewhere, such as disconnected ways or oneway arrows going head-to-head or back-to-back? Can you download the region around your route and compile it with mkgmap with logging enabled? Does it complain anything relevant? I do not travel by car that often. A year or two ago I remember a similar situation where a split way was merged. There was some error in the map data at that point, but I cannot remember what it was and cannot deduce it from the history of the ways. FWIW, it probably was node 27536320 at lat=60.4558175, lon=24.8496977, and I was heading west. Marko
On 2:59 PM, Marko Mäkelä wrote:
Right, that's why I asked where you were heading to and what the routing suggestion looked like. Was the Garmin unit configured to auto-recalculate, and was there some bad coverage (tunnels) that caused it to "jump" somewhere else and back? Or could there be some error in the map data elsewhere, such as disconnected ways or oneway arrows going head-to-head or back-to-back? I didn't see anything on a quick scan of the data all the way south to my destination. All nodes were connected. I usually rely on KeepRight to flag errors in the data and nothing is reported as far as I can tell. No tunnels, nothing, just a straight road to destination. MapSource doesn't seem to have the issue which is strange. Can you download the region around your route and compile it with mkgmap with logging enabled? Does it complain anything relevant? I will try it tomorrow if I have time. I do not travel by car that often. A year or two ago I remember a similar situation where a split way was merged. There was some error in the map data at that point, but I cannot remember what it was and cannot deduce it from the history of the ways. FWIW, it probably was node 27536320 at lat=60.4558175, lon=24.8496977, and I was heading west. Thanks. I am out of ideas given that MapSource works.
Francisco
On 2:59 PM, Francisco Moraes wrote:
Can you download the region around your route and compile it with mkgmap with logging enabled? Does it complain anything relevant? I will try it tomorrow if I hae time. I did this yesterday and generated a map and didn't see any issues on the mkgmap log. Mapsource routing works fine. I tried to test it on the GPS, but the move to location didn't appear to be working, so I can't verify the problem happens on the compiled map. I've uploaded it here: http://files.mkgmap.org.uk/download/48/gmapsupp.img
Francisco
On Wed, Jan 11, 2012 at 10:02:16AM -0500, Francisco Moraes wrote:
I did this yesterday and generated a map and didn't see any issues on the mkgmap log. Mapsource routing works fine. I tried to test it on the GPS, but the move to location didn't appear to be working, so I can't verify the problem happens on the compiled map.
I only have experience of the Edge 705. If I power it up indoors so that it will not get a GPS fix, it will give up after a few minutes and pop up a menu. One of the options allows me to set the current location. After doing that, I can set the unit to demo mode and calculate routes from the chosen location. Would this work for you? Marko
On Oregon you can force it with setting GPS-mode to GPS-simulation. Now you can route to any point and the Oregon can jump to that point or simulate the route. Henning
On Fri, Jan 06, 2012 at 06:05:06PM -0500, Francisco Moraes wrote:
I have seen some strange routing issues with the way below. If I am driving south, my GPS wants me to make a U-turn where the divided highway ends. It seems like the ways are not connected but I checked and I can't find anything wrong in the OSM data.
http://www.openstreetmap.org/browse/way/37649065
Any ideas?
One idea is the observation that in the USA (at least California), it seems to be common to have compulsory U-turns. On a divided street, the divider would not be interrupted in every crossing, like it usually is in Europe. But in this case, the Hopson Rd/Louis Stephens Road crossing seems to be fully connected both in OSM and in the Bing imagery. Was your destination to the east? What is the complete calculated route? There is a no_u_turn restriction at the south end of the way, added by aighes today. Is this your test if the turn restriction makes any difference? I guess you might want to add no_u_turn restrictions to the crossing as well, because the Bing imagery seems to suggest no_u_turn there. Note that mkgmap currently does not support ways in role=via. Best regards, Marko
On 2:59 PM, Marko Mäkelä wrote:
One idea is the observation that in the USA (at least California), it seems to be common to have compulsory U-turns. On a divided street, the divider would not be interrupted in every crossing, like it usually is in Europe. But in this case, the Hopson Rd/Louis Stephens Road crossing seems to be fully connected both in OSM and in the Bing imagery.
Was your destination to the east? What is the complete calculated route?
The destination was on Louis Stephens Rd south of the intersection. This was the second time I noticed this but I never found an issue in the data itself.
There is a no_u_turn restriction at the south end of the way, added by aighes today. Is this your test if the turn restriction makes any difference? I guess you might want to add no_u_turn restrictions to the crossing as well, because the Bing imagery seems to suggest no_u_turn there. Note that mkgmap currently does not support ways in role=via. Not sure what you meant here. I don't understand why I get this strange behavior just on this way and not on similar ones. I can look up others that are the same where a straight route works fine but I don't recall it right at this moment.
Francisco
participants (4)
-
aighes -
Francisco Moraes -
Henning Scholland
-
Marko Mäkelä