[Routing] Journey time prediction/traffic flow

[Stefan de Konink]

camaj schreef:
> --- Stefan de Konink <skinkie at xs4all.nl> wrote:
> 
>> camaj schreef:
>>> --- Stefan de Konink <skinkie at xs4all.nl> wrote:
>>>
>>>> Isn't this just the implementation of a dynamic
>> cost
>>>> function?
>>> Maybe, I haven't the foggiest idea what a "dynamic
>>> cost function" is!
>>
>>
>>
> http://www.google.nl/search?q=dijkstra+dynamic+cost+function&ie=utf-8&oe=utf-8&aq=t&rls=org.mozilla:en-US:official&client=firefox-a
>>
> Ok, well I haven't got the time to read all of that
> but from what I can gather (it's geared to a
> mathematically literate audience) it's essentially
> what I'm talking about. However the software, as I
> understand it, already exists, it's just not part of
> OSM. 
> 
> Any use of an algorithm like this is going to be the
> same whether your measuring time or distance
> 
> I notice Google maps has a very similar tool, but only
> in the US (which is why I didn't mention it) and only
> as an overlay (doesn't use during routing). In fact
> I'd already seen a similar thing for the UK on, I
> think, the highways agency website. Both limited
> themselves to motorways though.
> 
> There's also a traffic system on Microsoft live maps
> but either they've removed it or it's only accessible
> to US visitors.
> 
> This NY Times article discusses their methods and
> they're similar to what I proposed
> http://www.nytimes.com/2008/04/10/technology/10maps.html?_r=1&oref=slogin

Basically for every way/node a cost function based on historical data. 
If you route over the node you use Dijkstra/A* and let the cost or 
heuristic aspect to take in account the time you will arrive on that 
place. (With a nice error bound)


Stefan