More accurate route time prediction

[ah be]

As on some roads one may never achieve the standard predicted speed for a its type my idea is that one can record one's track and have a real vs predicted speed profile side by side, which allows to identify the highest deviations and thereby one could adjust the driving speed settings to get a more accurate routing performance.

This could be a two step system:

A)

• navit does a general comparison on what speed was achieved against the expected speed on a per highway-tag type.

eg. highway average real 93.8km/h vs highway average predicted 120km/h

• thereafter navit averages the results of the last n-routes and slowly adjusts the predicted speed for each route on a base of a marching average

B)

• Same as A) but instead of changing the global speed for a certain route type it changes the local speed for a segment (as I have no idea on how routing is done I don't know if one can apply personal patches on a specific street segment (50km/h -> 43.2 as there is always traffic; also I have no idea if such patches would work if one updates the map...)

[tryagain]

(B) Already can be done manually by editing navit.xml.

C) Divide amount of time needed to drive over previous way part by predicted time. Then use this coefficient to adjust predicted time. We also can have a number of such coefficients which will reflect actial and predicted time difference on the whole route and on some of its recent parts.

[ah be]

I was aware that I can adjust the speed for each highway type in the navit.xml file, but how do I get a table with highway type and average speeds per type for the trip?

[usul]

I'm not sure if we can really acomplish such a solution. A few notes:

• collecting a mass of data (FCD) is beyond that navit can/should offer
• profiling the driver can be a privacy problem