-
Notifications
You must be signed in to change notification settings - Fork 1
Building a routable network
Jip Claassens edited this page Aug 3, 2026
·
3 revisions
(outline, to be written out)
A geographic file of road geometry is not yet a graph. This page will describe the step from downloaded source data (Network data sources) to a connected, directed network that Dijkstra can traverse.
- From arcs to links and nodes: splitting polylines at intersections, building the node set, and the
F1andF2from node and to node convention used throughout this wiki - Planarity and grade separation: crossings that must connect versus bridges and tunnels that must not, and why naive intersection of geometry produces routes through viaducts
- Directionality: one way streets, and representing them as directed links
- Turn restrictions and turn penalties: where they matter, as in urban car routing, and where they can be ignored
- Mode specific subnetworks: filtering links by access rights per mode, and why a pedestrian network is not a subset of a car network
- Connectivity checking: finding disconnected components with connected_parts, and deciding which are genuine islands and which are data errors
- Barriers and crossings: level crossings, ferries and gates
- Network optimisation algorithm, reducing the network once it is built
- Travel speeds per mode, assigning impedance to the links
- Connecting origins and destinations to the network
- GeoDMS network functions
Accessibility modelling documentation
© Object Vision BV

Concepts
Networks
- Network data sources
- OSM
- NWB
- GTFS
- Building a routable network
- Network optimisation algorithm
- Travel speeds per mode
Origins & destinations
Computation
- Shortest path and the Dijkstra algorithm
- Origin Destination matrix (trip table)
- Isochrones and catchment areas
- Public transport routing
Indicators
Applications
Network models
Reference