Skip to content

Latest commit

 

History

History
154 lines (114 loc) · 3.71 KB

osrm_firstgo.md

File metadata and controls

154 lines (114 loc) · 3.71 KB
Raw

OSRM first go

This focuses on setting a basic local OSRM routing server using the default bicycle profile. Anyone is free to go further to potentially explore the modification of profiles.

This was done on MacOS.

Set up local OSRM

First, we need to set up a local server from shell. The references can be found here.

There are some dependencies that should be installed first. I used brew, which make it way easier.

# pre-requisites
brew install cmake boost boost-python3 tbb libstxxl libxml2 bzip2 lua luajit libzip openssl pkg-config

Then, we clone the OSRM repo and build the software using a PBF file for Leeds.

# Clone osrm-backend from git
git clone https://github.com/Project-OSRM/osrm-backend.git
cd osrm-backend

# Make
mkdir -p build
cd build
export TBB_DIR=/usr/local/opt/tbb

cmake .. -DTBB_DIR=$TBB_DIR
make

# Get the road data for Leeds
cd ..
wget -O leeds.osm.pbf https://download.bbbike.org/osm/extract/planet_-1.6757,53.7141_-1.4287,53.879.osm.pbf

# extract the data with bicycle profile
./build/osrm-extract -p profiles/bicycle.lua leeds.osm.pbf

# partition
./build/osrm-partition leeds.osrm
./build/osrm-customize leeds.osrm

Finally, we run the instance.

# Run instance
./build/osrm-routed --algorithm mld --port 5001 leeds.osrm --max-table-size 100000

With the instance active we go to R and specify the local host on osrm.

Excecute from R

So, we set the local host.

# Set environment variable to avoid fork safety issues on macOS
Sys.setenv(OBJC_DISABLE_INITIALIZE_FORK_SAFETY = "YES")

library(osrm)
Data: (c) OpenStreetMap contributors, ODbL 1.0 - http://www.openstreetmap.org/copyright

Routing: OSRM - http://project-osrm.org/

library(sf)
Linking to GEOS 3.11.0, GDAL 3.5.3, PROJ 9.1.0; sf_use_s2() is TRUE

# Set local host
options(osrm.server = "http://localhost:5001/")

For this example, we use the OD data generated in the README file. So, we read the data directly.

# Read OD points
od_geo <- sf::st_read("input_data/od_data_100_sf.geojson")
Reading layer `od_data_100_sf' from data source 
  `/Users/rafa/Desktop/routingday/input_data/od_data_100_sf.geojson' 
  using driver `GeoJSON'
Simple feature collection with 100 features and 3 fields
Geometry type: LINESTRING
Dimension:     XY
Bounding box:  xmin: -1.743949 ymin: 53.72819 xmax: -1.346497 ymax: 53.92906
Geodetic CRS:  WGS 84

Next, I prepare the OD into a suitable format to pass it to OSRM.

# Define OD points
origins <- lwgeom::st_startpoint(od_geo) 
origins <- sf::st_coordinates(origins)

destinations <- lwgeom::st_endpoint(od_geo) 
destinations <- sf::st_coordinates(destinations)

Routing from local OSRM server

Now, to keep the SF geometries together with the distance and duration estimates, I run one-to-one OD pair (parallel is causing troube in MD environment). I am not sure if this is the best way to do it. Given that OSRM has some multiprocessing capabilities internally, but so far, I am not sure how to implement it from R while preserving all the information.

# Run OD routes
system.time({
  osrm_routes <- lapply(1:nrow(origins), function(i) {
    route <- osrm::osrmRoute(
      src = origins[i,], 
      dst = destinations[i,], 
      overview = "full",
      osrm.profile = 'bicycle'
    )
    return(route)
  })
})
   user  system elapsed 
  0.215   0.009   1.116 

# bind rows
  osrm_routes <- dplyr::bind_rows(osrm_routes)

This takes about 1.4 in the quarto environment. However, in a parallel processing is reduced to about 1 sec.

plot(osrm_routes[,3])