Gravnav

maximize burritos

You can play around with GravNav at ommehta16.github.io/gravnav: just click two points, wait for a route, then explore!

What is GravNav?

A little project I've been working on!

Uses A* graph traversal, the OpenStreetMap Overpass API, express.js, hopes, dreams, and an INCREDIBLE amount of duct tape to show you driving routes that get "subtly" attracted to nearby Chipotles

Running GravNav

As is, you just need to open index.html in your browser of choice. By default, the app's configured to use my droplet (api.gravnav.ommehta.us) to find, process, cache, and send map data.

If opening index.html doesn't cause chunks to start loading, it's likely because your browser/computer is being silly (or just has extra, non-standard protections in place for straight html files!).

In this case, I've included silly-server.py. If you'd like to run it this way, ensure that you have Python>3.13. Then run:

python3 -m pip install flask
python3 silly-server.py

Then you should be able to load gravnav on localhost:3000

Self-hosting

The backend is just a single, simple(ish) express.js server. I have it set up on a 2GB DigitalOcean droplet, but really any machine with >1GB of RAM should be able to handle it.

To run it, ensure you have node.js>24 and npm>11.8.0 installed, then

npm i; node server.js;

You can then quit the server with ^C, and force quit it with another ^C!

If you'd like to run it in a production-y environment, I'd recommend using pm2. Check out their docs for more information on using it!

You can also get set up with a remote proxy! The nginx config for the server is in /nginx/gravnav.conf; just include it in your main nginx.conf file, change it to your needs, start nginx, and you should be good to go!

---

In any case, though, if you're self-hosting, you'll need to change the backend URL in getdata.js. Just set OSM_API_URL to the url of your server, and you're good to go!

How does it work?

On the server side,

• Get road + Chipotle data from OSM OverpassAPI for new tiles
• Simplify road graph (--> ~10% of nodes)
• Store road graph

When the client is initially loading, it:

• Floodfills to find the order to load chunks in
• Fetches chunk data from server
• Merges new data from chunk with existing data (resolve common nodes/intersection conflicts)
• Calculate new timespans for each road segment

And after we've gotten user input,

• Find the closest node to the click point
• Run regular A* on the graph for the closest route
• Run A* with modified potentials for the "Chipotle-est" route

Thanks for reading this far! This project has been so fun to write over the past month-and-a-half :)