Taxicab

When routing between two points of longitude and latitude, the built in routing functionality in OSMnx will find the nearest network nodes and route between those. This assumption is fine, and works for many applications but when you need routing with a little more accuracy you'll want to consider using Taxicab. Below are a few examples which highlight Taxicab use cases:

When the nearest nodes are not that close:

When routing along a single edge:

When routing along short routes:

When the nearest nodes are the same:

Installation

The easiest way to use Taxicab is probably via a PyPi install:

pip install taxicab

You can also install directly from this repo:

pip install git+https://github.com/nathanrooy/taxicab

Example usage

Taxicab is designed to be used as a drop-in replacement for the standard routing functionality found on OSMnx. So, like usual, download a portion of the OpenStreetMap graph:

from osmnx import graph_from_bbox
xmin, xmax = -84.323, -84.305
ymin, ymax =  39.084,  39.092
G = graph_from_bbox([xmin, ymin, xmax, ymax], network_type='drive', simplify=True)

Now, specify your origin and destination:

orig = (39.0871, -84.3105)
dest = (39.0880, -84.3200) 

Compute the route via the following:

import taxicab as tc
route = tc.distance.shortest_path(G, orig, dest)

Which can then be plotted:

tc.plot.plot_graph_route(G, route)

The returned route is a named tuple comprised of four elements:

• Route length in meters (or whatever your graph units are).

>>> route.length
669.0529395595279

• List of node IDs representing the bulk of the route (this is identical to OSMnx).

>>> route.nodes
[197546973, 2090608743, 197656382, 197633479]

• And two partial edges represented by shapely.geometry.linestring.LineString objects. If populated, these represent the first and last segments of the route that extend from the first or last node to some point along that edge.

>>> route.orig_edge, route.dest_edge
(<shapely.geometry.linestring.LineString at 0x7f1aa08067c0>,
 <shapely.geometry.linestring.LineString at 0x7f1a3ccbd580>)

User reference

taxicab.distance.shortest_route(G, orig, dest)

Parameters:

• G : (networkx.MultiDiGraph) - input graph
• orig : (tuple) - a (lat, lng) or (y, x) point
• dest : (tuple) - a (lat, lng) or (y, x) point

Returns: (named tuple)

• route.length : float - distance in meters of computed route.
• route.nodes : path - list of node IDs constituting the shortest path (this is identical to routes found in OSMnx).
• route.orig_edge : shapely.geometry.linestring.LineString - a partial edge representing the first non-complete edge in the route.
• route.dest_edge : shapely.geometry.linestring.LineString - a partial edge representing the last non-complete edge in the route.
• Note that if a route is successfully computed the distance will always be returned. However, depending on the length of the route and the underlying network, route.nodes, route.orig_edge, or route.dest_edge, may be null.

taxicab.plot.plot_graph_route()

Used exactly the same way as osmnx.plot.plot_graph_route except that it uses the route (named tuple) produced by Taxicab instead. See OSMnx function reference [here]