Trying to predict speeds theoretically is always going to be very error prone - you could imagine trying to get very clever and keep track of gradients and a sort of 'sinuosity' (or bendiness as I prefer to call it), but I'm not convinced it will be that reliable - you will have to put some big uncertainties on it.
I think it might be interesting to collect actual journey times by asking people to submit GPX tracklogs (and tell us what sort of vehicle they were using) - you could split the track log up to 'segments' between junctions, then calculate the journey time for each road segment for different vehicle types. It would take a bit of setting up to encourage people to submit the track logs, but it would be fairly easy to automate the calculation and collect the statistics. You could even monitor journey times over a long period to see if they are getting longer or shorter as the roads get busier.... I imagine this dataset being something separate to the main OSM data - it could be a relation for each road segment that could be picked up by routers from somewhere else. Herman Krauss has done something similar (but for altitude profiles) for his Google Summer of Code project, so the code to produce the relations exists, we would just need to link it to the GPX track log database to do the calculations. Graham. 2009/8/10 John Smith <delta_foxt...@yahoo.com> > > --- On Mon, 10/8/09, Lambertus <o...@na1400.info> wrote: > > > Yes, but as they are often represented as a node in OSM, > > giving them an average speed is meaninless. A time or point > > penalty seems more suitable. > > It works out to be 1.5s of lost time for a speed hump, traffic lights are > harder to predict if you're lucky you'll get a good run of them, unlucky and > you might have 3-5min per traffic light. > > > > > _______________________________________________ > dev mailing list > dev@openstreetmap.org > http://lists.openstreetmap.org/listinfo/dev > -- Dr. Graham Jones Hartlepool, UK email: grahamjones...@gmail.com
_______________________________________________ dev mailing list dev@openstreetmap.org http://lists.openstreetmap.org/listinfo/dev