[OSM-talk] [Talk-us] Am I mapping this wrong, or should the router be fixed for this?

[Maarten Deen]

It is more than sufficient for a time calculation to use the maximum 
speed, multiplied by some factor (smaller than 1), or even a fixed speed 
per road class. My car navigation has this (there are three speeds I can 
set) and usually the time is correct within a few minutes.
Much better is virtually impossible to achieve since you don't know how 
much traffic there is on the road so you can not predict waiting times 
at traffic lights or junctions.

But again: the duration of a route has nothing to do with the actual 
route calculation. You first calculate the route based on cost factors 
and then calculate the time you need based on speed profiles.

Maarten

On 2015-07-30 21:24, Colin Smale wrote:
> I assume you are talking about typical speeds, and not a practical
> maximum. A max speed will almost never be achieved, by definition
> actually as the vehicle speeds will have a certain distribution. The
> highest recorded speed will be the de facto practical maximum,
> assuming the driver survived.
> Routers could take account of hundreds of variables in their
> calculation of predicted journey time from A to B, but in practice
> their calculations make assumptions for most of them. For example,
> most of them assume the vehicle is a car, that it is technically not
> limited to any particular speed, that the weather is perfect, that it
> is daytime, that the driver is not "inexperienced". And then there are
> the other volatile variables like traffic density, road works,
> oversize loads getting in the way etc.
> Routers cannot take everything into account (this would preclude a lot
> of preprocessing to simplify the real-time calculations), so they use
> heuristics which work most often.
> So how would you define the concept of "typical speed"?
> --colin
> 
> On 30 July 2015 20:38:32 CEST, Richard <ricoz.osm at gmail.com> wrote:
> 
>> On Thu, Jul 30, 2015 at 08:00:55PM +0200, Colin Smale wrote:
>>> Practical maxspeed is useless as well. A straight wide road may be
>>> capable of hosting land speed records, but traffic density is
>>> likely to be a far more important factor.
>> 
>> yes, and this is what practical maxspeed is good for. Not
>> an ideal solution but works.
>> 
>> Richard
> 
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk