Talk:Average speed per way

My tracks won't be much use

...at least not in residential areas.

If I'm surveying by car in residential areas I stop to photograph street signs, postboxes, bus stops, etc. I sometimes do this on non-residential roads as well where it is safe to do so, particularly if I want to mark where speed limits change (by photographing the sign).

And as you point out different vehicle types will produce different results. Some roads I've walked, and that will bring the average way down (it could have been a traffic jam at walking pace - how would you tell?).

Good luck though. --EdLoach 08:15, 12 February 2009 (UTC)

We arrived at the same conclusion in the routing-mailing-list last year. The only usable tracks aould be the ones explicitely tagged for this kind of analysis by a yet to be defined tagging-schema. --MarcusWolschon 09:08, 12 February 2009 (UTC)

Ok I see some points there. But I think there will be some ways where a solution is possible (motorways etc.). Also statistical methods could be used to eliminate mappers :-) (that's what I meant with qualifying tracks). Of course we need a substantial number of tracks then. I think this is a huge task here but I am challenged somehow... --Gary68 12:00, 12 February 2009 (UTC)

wrong aproach

I think your aproach of adding such data to the OpenStreetMap is flawed. (On speedcollector.comyr.com I'm playing around with an idea of a simple database that collects real-world speeds.)

I would more like to: [*]collect the speed a real vehicle has on average on a highway=secondary,... instead of guessing such values [*]collect how fast e.g. "a medium speed car has on road XYZ on a friday afternoon"

But here you:

• You cannot collect data there to build an average. If so you would have to save not an attribute "avgspeed:(car-type)" but "allspeeds:(car-type):(0-1000000)" and this is insane.
• The tagging-schema is extremely limiting. You cannot combine few meassurements on a specific road with many meassurements on all roads of this type at this kind of time for this kind of vehicle.

I started a thread to discuss details here (less limiting then a wiki:talk -page and does not disturb thousands of users like the osm-mailing-lists).

--MarcusWolschon 14:15, 13 February 2009 (UTC)

indeed i think it'd be better to do this in the used software instead of tagging in OSM.. a routing software will be able to deduct this from the max-speed (for car navigation.. pretty useless for walking anyway unless you intend to do jogging in a busy city street). deducting could use simple algorithms like 80% of max-speed on highways, 60% on resident streets or diminish the 80% value by 30-40sec for every traffic light. --Blk 17:45, 16. Feb. 2009

Using rules like 80% of max-speed is no improvement and has nothing to do with the collection of real world speeds. It is not relevant here. (Please do not forget to sign your post)--MarcusWolschon 06:56, 17 February 2009 (UTC)

useful information could be how full the streets are at different times of day / weekdays.. > 100% being jams, 80% making lane switching more difficult but still reasonable speeds.. just ideas and wild guesses --Blk 17:45, 16. Feb. 2009

Automatic upload of traces by route-planning software

The most important thing at this point is to get traces. We should ask the route-planning software developers to ask their users to automatically upload anonymous traces (most likely with the Trackable option). Upload could be done under the user OSM account if available or under a generic OSM account for the software. So, Maps.me and OsmAnd developers could be asked to do so. Note that Navit and MapFactor Navigator Free also use OSM maps, but since they mostly rely on selling commercial maps, they most likely don't want to make OSM maps too good. That's probably why only Maps.me and OsmAnd let the user contribute to the maps. The RedBurn (talk) 08:24, 27 June 2018 (UTC)