search

graphhopper / map-matching

The map matching functionality is now located in the main repository https://github.com/graphhopper/graphhopper#map-matching
https://www.graphhopper.com/open-source/
Apache License 2.0
785 stars 274 forks source link

Improve matching for oneways #30

Open ghost opened 9 years ago

ghost commented 9 years ago

I'm trying to map match the following route - it is a series of lat-lon-time triplets taken at 1 Hz. screen shot 2015-07-01 at 11 41 11 am

After I put them through graphhopper's mapmatching software, I get the following output: screen shot 2015-07-01 at 11 41 12 am As you can see, there are a bunch of points on the right that are returned by graphhopper that are not in the route that was inputted. What I believe happened was a trip was detected to occur on the opposite end of the street, so graphhopper returned the closest shortest way to get from one side of the street to the other, hence the extra points. Is there anything I can do to reformat the data I input so this does not occur, otherwise what can I do to detect when this occurs (other than looking at its output)?

I'm also running into a problem with another route. The route I'm using looks like this: screen shot 2015-07-01 at 11 41 07 am

The points that I am getting back look like this: screen shot 2015-07-01 at 11 41 09 am I double checked the order in which points were inputted in my route - the first point is on the bottom left and continue on to the top right as they are inputted. The result that graphhopper is returning starts at the end of my trip (top right), and ends on the bottom left. I'm not sure why this is happening and I've tried reversing the input of my data and am getting the same result. Does anyone know why this is happening?

karussell commented 9 years ago

Thanks, for the detailed screenshots and description. In order to look more closely into it we would need some example GPX files that you use as input.

The last example looks like there are problems with one-ways in the road network but again: hard to debug if one doesn't have the coordinates.

ghost commented 9 years ago

Would it be possible for me to send you the GPX files? If so, what would be the best medium?

karussell commented 9 years ago

Best would be to use gpsies.com - there you can visualize them. Or use any file hoster

ghost commented 9 years ago

Thanks a bunch: Here are the tracks: http://www.gpsies.com/map.do?fileId=ypuvgrmbyqmyjnad http://www.gpsies.com/map.do?fileId=bqmpkljzzbijswhb

karussell commented 8 years ago

The first example works with the latest master now.

The second example seems to be in reverse order and produces indeed a wrong match (as the algorithm cannot follow the oneway in the wrong direction it produces artifacts). But on GPSies.com there is a 'Reverse track' option which improves the match but still not like intended I guess.

Will have a look when I improve other matching stuff.

rtrip

karussell commented 8 years ago

Related to #63

michaz commented 6 years ago

The second example seems to be in reverse order

What makes you think so? It even has time stamps.

karussell commented 6 years ago

What makes you think so? It even has time stamps.

Oh, this is a bit old and I cannot remember exactly. But likely it was against several of these oneways in the map (maybe they were wrong at that time and are fixed now, but you can still see the old status as there are a few arrows on the screenshot)

michaz commented 6 years ago

Maybe they were walking.