osrmIsochrone(): 'loc' is too far from the OSRM network but loc is not too far... about osrm

[gmachin11]

Hi!

I'm having the same issue, also i noticed that only happens when you add a negative coordinate to the loc requirement, if you type positive positive coordinates it works.

[rCarto]

@gmachin11 Thanks for the hint!

[gmachin11]

You're welcome, i tried for each combination of positive and negative, only worked for the positive positive one, like in here:

osrmIsochrone(loc = c(13.43,52.47), breaks = seq(0,12,2))

[rCarto]

I think that this problem comes from the demo servers (http://router.project-osrm.org/r & https://routing.openstreetmap.de/). It seems that they do not manage the point encoding in google polyline format very well.
I've tried the latest version of OSRM (v.5.27.1) with docker and it is working as expected.
The first priority of osrm package is to be up to date with the OSRM server latest version. So the package is ok in principle & this is not a bug 😄 .

Solutions:

1. Users affected by this problem could reach to OSRM demo server maintainers and see if an update of the OSRM version is possible from their side. According to their doc the routing.openstreetmap.de demo server uses v5.23.0 (released in Oct. 2020).
2. I want users to be able to test the package with demo servers before deciding to use their own instance, so I'll try to circumvent the problem within the package and release an update soon.

[rCarto]

Should be fixed in the dev version of the package.

remotes::install_github('riatelab/osrm')

[ClaudeGrasland]

Yes ! 👍
I have reinstalled the dev version and now it works.

In fact I had inverted the coordinates of latitude and longitude in my first example and it was the reason why it did not work (the point was in the middle of the sea). 😞

[ACouillet]

Good morning,
using R osrmIsochrone, I've got the same error : 'loc' is too far from the OSRM network.
No negative value in coordinates and dev version installed from Riate.
It seems like if sf points located at the ends of the roads couldn't be taken into account for isochrones calculation.

[rCarto]

@ACouillet
Thanks for the feedback, can you give me the coordinates of the point you're trying to get isochrones from ?

[ACouillet]

@rCarto
Merci Timothéé !
geometry (x / y, WGS84)
6.164979 / 44.11091

In fact, I deduced that among the 15000 points to be treated, some are connected to the network but on portions defined in OSM as roads not accessible to cars so logically the "car" isochrones cannot be calculated...

Also, in some cases, it seems that portions of the network are poorly defined (after verification with IGN data, they should accept an iso calculation which is still impossible).

It is therefore more a problem of characterization of the network in OSM than a malfunction of the program !