I tried running the route search engine (OSRM) easily with a container
background
Once again, I found it very useful to use Docker (container), so I decided to organize the case as a memorandum.
What did you want to do in the first place
I wanted to easily use ** OSRM (Open Source Routing Machine) **, a route search engine that uses ** OpenStreetMap (OSM) **, but it seems that it is troublesome to build the environment, so I can not do anything. I was investigating.
What, the already built environment was provided as a Docker image
In Project-OSRM / osrm-backend, there was the following description.
The easiest and quickest way to setup your own routing engine is to use Docker images we provide.
Moreover, just by executing the following command, the container image is acquired (downloaded) from dockerhub, and a container (as a route search engine) is created and started from that image.
docker run -t -i -p 5000:5000 -v "${PWD}:/data" osrm/osrm-backend osrm-routed --algorithm mld /data/kanto-latest.osrm &
In this example, ** osrm / osrm-backend ** is the container image and ** osrm-routed ** is the command to run inside the container.
However, the following work is required in advance for this execution. Among them, the following formatting and optimization processing is executed by starting a new container from osrm / osrm-backend (starting the command for that processing as a container).
--Download the map (OSM) used by OSRM --Formatted for use with OSRM --Optimization of speed for walking, cars, bicycles, etc. --Other
Originally, it is said that processing other than ** 2. 6. 7. is required ** by the following procedure, so you can see that you can try it faster and more reliably by using Docker.
- Update and security
- ** Download Map Export **
- Dependency installation
- Compile OSRM
- STXXL configuration
- ** Map extraction **
- ** Calculation of travel time **
- Running and testing web services
- Nginx setup
- Supervisor installation and configuration
So, I was able to reconfirm that it is very easy to build an execution environment by using a container image.
For reference, I executed the following command on the route search engine launched above to get the route information (JSON format).
In this example, in order to get the walking route from Kita-Kamakura to the Great Buddha of Kamakura, each location information is thrown to the search engine as a parameter.
curl --noproxy 127.0.0.1 'http://127.0.0.1:5000/route/v1/walking/139.545056,35.337103;139.535689,35.316696?alternatives=3&geometries=geojson' | jq 'def hexdec(i): "0123456789abcdef"[i:i+1]; {"type": "FeatureCollection", "features": [[.routes[].geometry] | [., keys] | transpose[] | {"geometry": .[0], "type": "Feature", "properties": {"stroke-width": 2, "stroke": ("#" + hexdec(15-.[1]*2) + hexdec(.[1]*2) + hexdec(.[1]*2))}}]}' > routes_kamakura.json
Then, when the acquired Json format file is loaded on ** geojson.io **, there are multiple walking courses from Kita-Kamakura to the Great Buddha of Kamakura as shown below. I got the route. (This is a separate article that I would like to organize)
Recommended Posts