problem with only_straight_on turn restriction

[paulcone]

I've loaded up OSM for Oregon (from geofabrik.de) into an FGDB and am doing some routing in downtown Portland with Network Analyst (10.2). I'm getting some results that I wouldn't expect -- it's doing a U-turn on to a parallel segment of the same (divided) road, even though I have U-Turns at Junctions set to Not Allowed. I think that Network Analyst is not seeing them as U-turns because they are different segments, and I'll accept that. But these intersections are also tagged with restriction=only_straight_on. From what I can tell by looking at the nd_turns, it looks like some turns that are not being imported properly from the OSM data -- specifically those tagged only_straight_on. I took a look at the code but I'm not really a programmer so I can't tell what's going on. Can someone explain the logic in how OSM turns are imported into a network dataset?

[eggwhites]

Hello Paul, thanks for reaching out and apologies for the delay, we've been ramping up for our Developers Summit in Palm Springs next week. So to your question regarding how the Create OSM Network Dataset tool defines turns - it uses a configuration file that you could definitely update per your specific FGDB. You browse to the config file when you run the tool - for example, \ArcGIS\Desktop10.2\ArcToolbox\Toolboxes\ND_ConfigFiles\DriveGeneric.xml. In this file, you'll note that a restriction for "only_straight_on" is not defined. You could add it into the file - make a copy of DriveGeneric.xml, and find the section where the Oneway is defined. Yes, I know its not the same as a restriction - you could also think about adjusting the TurnRestriction section. So in this section, you'd need to accommodate for "only_straight_on".

[eggwhites]

Hello Paul, The developer of the tool is standing here with me now, and said there's more to it than I had thought. The real issue is there is a known issue with how the tool handles turns that are only_straight_on with multiple points defining the turn. It essentially wants to round off the turn. So you have indeed found a bug - I'll tag this thread as such. Will you send me the bounding box of the area you are looking at, so we can easily get the data to recreate?

[paulcone]

Hi Christine,

Thanks for the prompt replies. I'm working with data in Portland, Oregon. For the two examples above, the first is the Hawthorne Bridge in Portland, which is min lat 45.5114296, max lat 45.5151285, min lon -122.6752281, max lon -122.6659584. The second example is Burnside Street just west of the river, which is min lat 45.5223828, max lat 45.5234953, min lon -122.679863, max lon -122.6767409. I have a few other examples also in the Portland area but hopefully those should give you enough to work with.

Paul

[eggwhites]

Thanks Paul!

[paulcone]

Hi Christine,

Any update on this? I am registered to attend the Routing workshop at FOSS4G on Tuesday, and I'm hoping this is fixed by then.

Paul

[eggwhites]

Hello Paul - we are looking at this, but I'm not sure if the fix is already committed in the code. I know it is not in the latest release of the ArcGIS Editor for OpenStreetMap installer. Will find out more and update you soon.

[paulcone]

I've noticed lots of work on this project. Any update on fixing this specific issue?

[robblis]

This is being looked at. No, solution yet.

From: Paul Cone [mailto:notifications@github.com] Sent: Friday, January 16, 2015 2:53 PM To: Esri/arcgis-osm-editor Subject: Re: [arcgis-osm-editor] problem with only_straight_on turn restriction (#40)

I've noticed lots of work on this project. Any update on fixing this specific issue?

— Reply to this email directly or view it on GitHubhttps://github.com/Esri/arcgis-osm-editor/issues/40#issuecomment-70335887.

[ThomasEmge]

Paul, I made changes in the code such that your outlined examples won't happen any more and turn restriction are properly build. However you would still need to expand our sample network model to properly handle a very complex Portland situation.

Thomas

[paulcone]

Thomas,

Can you explain more what you mean by "a very complex Portland situation"? How is it complex?

Paul

On Thu, Jan 22, 2015 at 12:25 PM, Thomas Emge notifications@github.com wrote:

Paul, I made changes in the code such that your outlined examples won't happen any more and turn restriction are properly build. However you would still need to expand our sample network model to properly handle a very complex Portland situation.

Thomas

— Reply to this email directly or view it on GitHub https://github.com/Esri/arcgis-osm-editor/issues/40#issuecomment-71092073 .

[ThomasEmge]

The challenge arises from the way the OSM data is structured and the way ArcGIS would define a network. Currently each vertex is a junction in the network and not all vertices/junctions have a turning restriction or not all edges have traffic directions defined. For example going up Northwest Broadway will eventually result in a U-turn when turn restrictions are no longer present and other roads attributes allow for such a maneuver. What I am trying to say to say is that our network definition suggestion called "GenericDrive.xml" doesn't completely model the Portland situation and our network instructions need refinement. We are looking for feedback on what is or would be the appropriate model.

[paulcone]

Thomas,

I'd like to help more on this. Can you give me some guidance on what sort of description you need for an appropriate model?

Paul