An overview of the components is given at http://sumo.dlr.de/wiki/Sumo_at_a_Glance There should be no need to dig deeply into the software design (but do ask again if more questions come up). All relevant code in in src/netbuild/NBHeightMapper (only used within NETCONVERT). The data is loaded in loadTiff() this is where the error originates. The z data is queried in getZ(). Here the coordinates are rounded rather than interpolating from the 4 nearest data points.
From our point of view it is very important to have tests for every change/extension of application behavior. (See http://sumo.dlr.de/wiki/Developer/Tests) regards, Jakob 2015-10-12 9:25 GMT+02:00 Matej KUBICKA <[email protected]>: > I will be implementing this anyway so I might try directly within > SUMO/NETCONVERT, if the overhead won't prove significant. Could you point > out where I can read about SUMO software design? It would help with > orientation in the code. > > Thanks, > ~Matej. > > Le 2015-10-12 06:54, Jakob Erdmann a écrit : > > SUMO currently does not interpolate the elevations. I understand the >> problem and there is now a ticket for that (see >> http://sumo.dlr.de/trac.wsgi/ticket/1962 [1]) >> >> >> regards, Jakob >> >> 2015-10-11 22:57 GMT+02:00 Matej KUBICKA <[email protected]>: >> Ok, so if my understanding is correct I can enforce segmentation of lanes >> at different elevation with --geometry.max-segment-length. With my >> elevation model having resolution 25x25 meters I can set this segmentation >> to, say, 12 meters, which will enforce that no elevation samples get >> skipped. >> >> Does SUMO interpolate the elevations, or not? If not, then the elevation >> profiles will contain artifacts which will have ugly effects on our >> simulations. >> >> For example, imagine that we set the segmentation per one meter. If you >> do not interpolate the data, then there will be mostly flat edges with a >> number of steep steps in elevation (effect, easily so steep that the car >> would essentially crashed in them :-)). >> >> ~Matej. >> >> PS: I will try to find a time to take a look at the GeoTIFF issue. >> >> Le 2015-10-11 21:37, Jakob Erdmann a écrit : >> >> Hello, elevation data may be added to all lane geometries. The netconvert >> option >> --geometry.max-segment-length can be used to control the spatial >> resolution (without the need to add any junctions). We haven't done a lot >> of testing with geotiff heightmaps. If you can provide a failing sample we >> may be able to track down the error. Please try to find the smallest sample >> possible. >> >> regards, Jakob >> >> 2015-10-11 16:36 GMT+02:00 Matej KUBICKA <[email protected]>: >> >> Hello, >> are elevation profiles incorporated in the shape of the lanes? - or, is >> it only junctions that have assigned elevation? >> >> I tried to add GeoTIFF height map to the Luxembourg scenario using "$ >> netconvert --sumo-net-file LuSTScenario.net.xml --heightmap.geotiff >> EUD_CP-DEMS_2500045000-AA.tif -o LuSTScenario.2.net.xml", hoping that >> diff would provide some clues. >> >> Its output: >> _Error: Unknown data type in EUD_CP-DEMS_2500045000-AA.tif._ >> _Warning: In lane '--32572_0': Could not build shape._ >> _Warning: In lane '-32572_0': Could not build shape._ >> _Quitting (on error)._ >> >> The error is pretty generic and the altitude data come from official >> source, so they should be fine. Can SUMO read the projection of the >> data? The dataset uses epsg::3035, Lamberth Azimuthal Equal Area >> projection. >> >> Thank you, >> ~Matej. >> >> -- >> >> Matej KUBICKA >> >> Phone: +33 1 69 85 17 81 [1] >> >> ------------------------------------------------------------------------------ >> _______________________________________________ >> sumo-user mailing list >> [email protected] >> https://lists.sourceforge.net/lists/listinfo/sumo-user [2] [2] >> > > Links: > ------ > [1] tel:%2B33%201%2069%2085%2017%2081 > [2] https://lists.sourceforge.net/lists/listinfo/sumo-user [2] > > Links: > ------ > [1] http://sumo.dlr.de/trac.wsgi/ticket/1962 > [2] https://lists.sourceforge.net/lists/listinfo/sumo-user > ------------------------------------------------------------------------------ _______________________________________________ sumo-user mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/sumo-user
