Nominatim VS Open Source Routing Machine (OSRM)

You can simply use geopy API to get longitude and latitude from address. geopy includes geocoder classes for the OpenStreetMap Nominatim, Google Geocoding API (V3), and many other geocoding services. The full list is available on the Geocoders doc section. Geocoder classes are located in geopy.geocoders. Here…

I have had a lot of "fun" with this feature in the past. It has earned a permanent place in my setup script:

ALTER SYSTEM SET jit=off;

Basically a performance blackhole for more complex queries. Most OSM related projects disable it straight away as it only creates headaches.

https://github.com/osm-search/Nominatim/pull/2559

Because osm.org and OsmAnd use different search engines. On osm.org it uses the Nominatim geocoder which is good at parsing addresses the way we are used to writing them. It is also running on a server with a lot of storage and memory resources. OsmAnd is built for offline use so it can't use Nominatim and has much more limited computing resources available to it. It is searching through the map files locally on your phone and however they have chosen to implement that search algorithm seems to perform poorly on addresses.

Ako skupa s adresama imaš i njihove lokacije možeš i koristiti gotov geocoder (e.g. Nominatim je free i koristi OSM podatke, možeš i deployati svoju instancu), pa iz rezultata naći geometrijski najbližu adresu koju imaš.

Then implement a use case. We are going to use Geocoding use case with Nominatim provider. But you can, of course, use a different provider. Copy the example code into your server.js file and make a few changes, so we can send the information we will receive from the user:

There are reasons why this tech was developed: because GPS can take a long time to sync up to the satellites, because GPS often doesn't work indoors, because your phone can keep a local cache of locations, and so you have the ability to find places by name or address (Geocoding) using nominatim ( https://nominatim.org )

and nominatim which does the geocoding/reverse geocoding: https://nominatim.org/

Then it depends on how the location data is conveyed in the group chat. If it's a text message you need to parse it and possibly use a mapping service to do geocoding. Eg. convert "22 Main Street" or "upper Brooklyn" into GPS coordinates. Besides Google Maps API you can maybe use OpenStreetMaps via Nominatim. If it's something more like "we just passed John's house" then you need to figure out how to write some logic to determine a location from that. Or maybe they just used the WhatsApp location share feature, then you probably don't need mapping at all.

then called the Open Streetmap api described here using Power query: https://github.com/Project-OSRM/osrm-backend/blob/master/docs/http.md

OSRM: A super fast and easy to use routing engine that runs on OSM data. You only need to run 5 lines of code to (1) download a .pbf from Geofabrik, (2-5) download the OSRM docker image and pre-process the OSM data. There are also 3 profiles predefined that you can use: car, bike, foot (e.g. foot.lua). It basically hosts a local server. I find the easiest way is to combine it with the osrm R package. I have seen you also need to adjust for the elevation. I think I have seen some custom LUA profiles that also account for DTM derived elevation changes as an additional weight.

I am trying (and failing) to setup an OSRM instance with docker compose. The issue I've encountered is with setting up osrm-backend. According to the Github and Docker Hub pages, the process is as follows:

I am creating Rust bindings to https://github.com/Project-OSRM/osrm-backend using autocxx. This allows for fast routing on openstreetmap data. They have a nodejs server, but I'm going to write a Rust one with these new bindings. I am finding that many of the config classes in C++ are't Plain-Old-Data structures, so are opaque structs, so I have to write getters and setters in C++.

> As a cyclist I’m almost always disappointed by google, apple, and Valhalla

Have you tried the OSRM bike config? (The one you find in https://github.com/Project-OSRM/osrm-backend/blob/master/pro...)

This is a 30 second exercise on OSRM, or any other service that provides directions.

As for the rules for routing algorithms, OSMR's foot profile looks like good place to start. You could copy the file into something like for_women.lua and extend it with rules for e.g. street lighting and any other features of interest. Of course, it would be best to reach out to the other devs first.

I'm trying to calculate walking distances using OSM and OSRM but this results in distance matrices that violate triangle inequality. This is a pretty bad issue for the spatial models I'm building. I was wondering whether there are alternatives to OSRM that guarantee that triangle inequality is respected. Mostly, I really don't care about the routing algorithm looking at red lights, whether streets are one or two way streets, private roads, etc. I just need the distance along pathways between points.