Ah man, I have been reviewing the bike profile, didn't see the 'weight_name' parameter examples in the car.lua...
Thx for the input, and your effort! I go explore some of your suggestions... On 7/11/19 2:09 PM, Frédéric Rodrigo wrote: > Le 11/07/2019 à 10:42, André Siefken a écrit : >> Thx Frédéric for the immediate reply, and sorry for me to take my time >> to respond. >> >> I will rewrite the profiles to test general soft restrictions, however >> my idea here is to have a profile that e.g. strictly uses distance >> between possible matched waypoints to get the overall match. > > Yes, you can active the distance as objective in the profile > > https://github.com/Project-OSRM/osrm-backend/blob/master/profiles/car.lua#L19 > > > >> How would the matching behave if I'd remove all road type based >> restrictions (e.g. vehicle type allowed or not, speed limit, oneways)? > > To have soft restriction you must keep it with high weight. > > > You can set this restricted road type into classes on profile, to > select them dynamically on request > > See > > exclude {class}[,{class}] Additive list of classes to avoid, > order does not matter. > > in > https://github.com/Project-OSRM/osrm-backend/blob/master/docs/http.md#general-options > > >> I do see where the algorithm needs certain base variables from the >> network to estimate a likely route, though, and do have problems seeing >> if this is feasonable at all, and I do ask instead of trying because it >> will take me some time to get familiar with profile changes... >> >> André >> >> On 7/9/19 10:14 PM, Frédéric Rodrigo wrote: >>> Le 09/07/2019 à 22:03, André Siefken a écrit : >>>> Hi @all, >>>> >>>> I'm exploring ways to have the matching algorithm work agnostic to >>>> (most) road restrictions and rather 'trust' the trace I pass in. >>>> >>>> I receive continuous GPS locations, with moderate to high >>>> resolution in >>>> time, from bikes as well as cabs or buses. It just so happens that >>>> some >>>> idiots ride their bikes on roads that OSM (and most laws) thinks they >>>> shouldn't, as well as some cars ignoring bus lane restrictions or >>>> buses >>>> leave their lanes. In those cases, matching fails one way or another >>>> ([No match], or only partial matches). >>>> >>>> I have a hard time wrapping my head around if a profile can actually >>>> fullfill restriction and/or even vehicle type (I can easily work with >>>> multiple graphs, though) agnostic matching, which likely implies a >>>> similar agnostic weighting. I imagine a shortest connection/path >>>> matching should do just that, but then I'm at a loss if that is >>>> actually >>>> the case and if so, how to generate such a profile. >>>> >>>> Your insights would be much appreciated... >>>> >>>> André >>> >>> I think the weighting is the right think to do. Use high weight on way >>> there is no legal access. It's some kind of soft access restriction. >>> You can even weight the opposite of a one way restriction. >>> >>> But you will have to rework all the samples profiles provided in OSRM. >>> >>> >>> Frédéric. >>> >>> > -- pgp-key attached
0x0024705C4FC20AF6.asc
Description: application/pgp-keys
_______________________________________________ OSRM-talk mailing list OSRM-talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/osrm-talk