I managed to delete my other post by mistake and apparently Discourse doesn’t show a dialog to prevent that. Sorry for the double post.
Hello. I’m working with some very poor GPS data sets (unfortunately) from street cleaning vehicles and I need to match the GPS traces to the map so I can calculate how much of the street was actually cleaned. I’ve managed to get some satisfactory matches, but I found out that the snapped_waypoints is always empty no matter what I try.
Is there any way I could correlate the output with the input?
My dataset also includes information on the status of the cleaning brushes (if they’re running or not) but it’s not reliable to do matching only where cleaning because (especially in tunnels) the GPS is pretty bad and on shorter sections the GPS traces can be quite wobbly resulting in bad matching, but if I try to match the whole dataset it very nicely detects when going through tunnels or over bridges.
I’ve already tried cleaning up my dataset by removing outliers, impossible segments or prolonged stationary periods (also tried some simplification algorithms but found it pretty unreliable especially do to the many u-turns these vehicles make).
Unfortunately I can’t upload files but here’s an example of the GPS traces I’m working with. There are many places where the traces are fine, but also many examples like the one in the lower half where the GPS signal was extremely poor.
