NYC Open Routing

Routing in NYC with pgRouting and authoritative NYC data. This project is currently a proof of concept and not intended for real world routing scenarios.

Features

• Driving, Walking and Biking routes using DCP's LION data
• Address searching using Geosupport and geosupport-suggest
• Routing capabilities using pgRouting
• Web API using Flask RESTful
• Frontend using React JS and Mapbox GL JS
• Containerized with Docker

Getting Started

This projected is orchestrated with docker-compose.

1. Clone this repo and cd into it:

   git clone https://github.com/ishiland/nyc-open-routing.git

2. Build the project.

   docker-compose build

3. Start the project.

   docker-compose up -d

   If this is the first time starting, the api container will download and install geosupport. You can upgrade geosupport by specifying a different version in the .env file and restarting the service.

4. Import the Lion data using the following command:

   docker-compose exec api sh /data-imports/scripts/import-lion.sh

   You can also specify a version of Lion:

   docker-compose exec api sh /data-imports/scripts/import-lion.sh 23a

5. When its complete navigate to http://localhost:3001

api

Use the flask api to query routes directly. All successful requests return GeoJSON w/MultiLineString geometries. You can comment out the client container in the docker-compose.yml if all you require is the api.

Parameters:

• orig: Origin coordinates. Expects a comma separated lat,long.
• dest: Destination coordinates. Expects a comma separated lat,long.
• mode: Travel mode. Can be drive, walk or bike

Example request: http://localhost:5001/api/route?orig=-74.0117,40.649221&dest=-73.951458,40.797061&mode=drive

response:

{
    "features": [
        {
            "properties": {
                "seq": 1,  // sequence of segment in route
                "street": "3 AVENUE",
                "distance": 260.679437746423,  // distance of segment in feet
                "travel_time": 0.0987422112675845,  // travel time of segment in minutes 
            },
            "geometry": {
                "type": "GeometryCollection",
                "geometries": [
                    {
                        "type": "MultiLineString",
                        "coordinates": [
                            [
                                [
                                    -74.01149956053376,
                                    40.65037686591395
                                ],
                                [
                                    -74.01207536260924,
                                    40.649811516367066
                                ]
                            ]
                        ]
                    }
                ]
            }
        },
...

Further Ideas

• Incorporate traffic data to more accurately calculate driving costs. Live or static traffic data would be a significant improvement.
• Add public transit modes using MTA data.
• Ferry schedules for more accurate biking/walking cost.
• Travel time isochrones for selected addresses.
• Line Merge segments where possible to optimize and reduce complexity of graph.
• Optimize functions by adding BBOX parameter.
• Add turn by turn directions with angle (left, right, sharp right, etc. ) similar to google maps.
• Send GeoJSON directly from the database using ST_AsGeoJSON to remove gdal dependency in api container.

Resources

• https://github.com/mixedbredie/highways-for-pgrouting
• https://docs.pgrouting.org/latest/en/pgRouting-concepts.html