Hi, On 10/24/2015 12:46 AM, Greg Troxel wrote: >> 1) creating a table listing each street on his road list to an array of OSM >> Way IDs, and somehow indicating to the algorithm that those ways are >> preferred. But this would require constant maintenance as people >> edit/split/merge/erase ways. >> >> 2) considering the street names, and somehow making the routing algorithm >> prefer those ways. > > The basic issue is keeping two databases in somewhat sync. If there is > KML, I would be inclined to have a combining step where the preferred > roads are matched up with OSM data by geometry and then the OSM ways > marked with the special tag (locally). Here, any of id, name, or > shape/location could work. shape/location is probably pretty stable.
I believe that both major OSM routing engines, OSRM and Graphhopper, have some way of track matching where you can give the algorithm a track (i.e. your KML) and it will compute the most likely sequence of OSM ways followed to achieve that track. OSRM at least also has a way to bring in external data into the graph building by way of LUA scripts, so this could be used to apply the weighting to those segments identified by the matching step. Bye Frederik -- Frederik Ramm ## eMail [email protected] ## N49°00'09" E008°23'33" _______________________________________________ Routing mailing list [email protected] https://lists.openstreetmap.org/listinfo/routing
