> I've done a few tests with the online demo of their contraction > hierarchies routing engine and the results are astonishing: South- to > North- Germany in 65 milliseconds. Very impressive. But I don't know how > many parameters are considered on each edge (turn restrictions, traffic > lights, sharp corners etc).
Did you try that one ? http://algo2.iti.kit.edu/schultes/hwy/demo/applet.html As far as I know, the online demo is running the Highway Hierarchies algorithm. But still, most of the 65ms is used for communication. With the current Contraction Hierarchy code it should be even less. > It would be awesome if this routing engine could be tested with OSM > data. I.e. how much disk space, preprocessing time and memory needed for > generating a routing database containing the whole OSM database, and > also how these parameters look like when doing routing calculations on > that database. I experimented on the road network of Germany about a year ago. Preprocessing was something like 5 mins or so. Have a look at the Contraction Hierarchies paper at https://algo2.iti.uni-karlsruhe.de/routeplanning.php It states the times for commercial european and north american road networks that are more dense than the osm data. --Dennis _______________________________________________ Routing mailing list [email protected] http://lists.openstreetmap.org/listinfo/routing
