bobdeng1974 / map_matching

Algorithms to find the streets that a vehicle should have traveled to generate a given GPS track

Map-Matching Algorithm

My implementation of the map matching algorithm from this article. The goal is to get the streets from a gps track.

This is how it looks like: The gray line is the gps trace and the colored lines describe the map-matched most-likely route in the streets for the vehicle.

For reference read the resulting article.

Setup

Install:

• python3
• redis >= 3.2.0

Download the data using the overpass api, there is a simple curl command to do that in the overpass/ folder:

$ cd overpass/
$ ./get_street_graph.sh

Install python dependencies. You may want to put them inside a virtualenv:

$ pip install -r requirements.py

Upload the street graph to redis:

$ ./main.py loaddata

Load the lua scripts:

$ ./main.py loadlua

run the mapmatching task with a geojson file with a single gps track and a number indicating how many points of the input gps track to process. Use a very hight number if you want the whole route processed

$ ./main.py mapmatching data/route.geojson 5