bicycle routing

[emiltin]

osrm should handle bike routing

[emiltin]

a good description of how cyclestreets.net imports, filters and proceses OSM data to handle bike routing in UK/Ireland: http://www.cyclestreets.net/journey/help/osmconversion/

[DennisOSRM]

Thanks. Feedback very much appreciated.

[DennisOSRM]

See also issue #10.

[emiltin]

how do i specify bike profile in the request to the server?

[DennisOSRM]

Emil, what do you think? How far are we along?

[emiltin]

i think we're doing very well, thank you for all your efforts!

to reach a first milestone of a useable bike router, i think what's mainly missing is some additional tag parsing, like the cycleway=opposite*, which i know this should be in the pipeline. i suspect there might be others, but it shouldn't be a big thing to implement.

looking beyond that, there are some things that would be very nice:

• cargobikes. shouldn't be that far away, but requires taking into account bollars etc.
• inclinations can be important for bicycles. i know you've been working on using the nasa data. using the inclination tag in osm would be an alternative. what's your plans/thoughts?
• turn penalties. i'm not sure wether this is already implemented?
• the whole time-based stuff like opening hours for parks and street lighting.
• option to route across streches where you can only push your bike - like intersections. osrm should probably handle bicycle=dismount.
• penalty for crossing intersections with traffic lights
• surface. choose ashpalt over cobbelstone..

but overall i think things are progressing nicely indeed. a lot of bugs have been squashed - i now have a significantly harder time finding bugs :-) and i actually get sensable routes most of the time.

[DennisOSRM]

The cycleway parsing is undergoing tests right now. It may arrive monday or tuesday.

Turn costs can be used in principle. All technical prerequisites are there. Only a thorough modelling of turn costs is lacking. Same holds for elevation measures which uses the NASA data. It's there but misses a good modelling of a height penalty. OSM tags are too unreliable for that, unfortunately. Also, traffic light penalties are coming sooner than later.

[emiltin]

sounds great!
a useful modelling should be forthcoming soon from the bikeability research project, done at copenhagen university. they've been doing both a stated preference study as well as analyzing gps tracks. based on this they will be able to provide weights for things like number of turns, street type, green index, etc. in other words, how much imoprtance does bicyclist put on each of these parameters when choosing a route?

regarding height tags in osm, why do you think they're unreliable? because they're not widely used?

[emiltin]

regarind traffic ights and turn penalties - i assume this would be settings in the speedprofile, where you can adjust the penalties? it would be interesting for me to experiment with this.

[emiltin]

regarding pushing bikes against oneways streets: #78

[emiltin]

i'm closing this, as it's overly broad. individual elements are covered in individual issues. several things are not yet implemented, like elevantion, separate speed/impedance and turn penalties.