Documentation on modifying cost function

Open Source Routing Engine
1

Hi everybody,
Is there any guide or documentation on how to use a specific cost function for the routing calculation?
My cost function should look like something like this:
for edge e linking e1 to e2
cost(edge e) = param1 * some_heuristic_distance(e2, fixed_point B) + param2 * some_probability(edge) + param3 * length(edge e)
Then I would like to route based on minimizing this specific cost.
I’m really lost in all the Java classes and can’t find a guide or documentation on what specifically should be modified to change the cost function. Any help would be highly appreciated. Thank you!

2

You can modify the code in CustomWeighting.calcEdgeWeight to do that and inject the new class with your own WeightingFactory or create a new class that implements the Weighting interface. See the SpeedWeighting for a very simple implementation.

3

Thank you very much for the reply.
I am still very lost. Please bear with me as I am new to Graphopper and Java dev too.
So I tried a top-down approach to see what needs to be changed.
In this simple example of route calculation:

class AnotherExample {

    public static void main(String[] args) {
        String relDir = args.length == 1 ? args[0] : "";
        GraphHopper hopper = createGraphHopperInstance("src/main/resources/berlin-latest.osm.pbf");
        routing(hopper);

        // release resources to properly shutdown or start a new instance
        hopper.close();
    }

    static GraphHopper createGraphHopperInstance(String ghLoc) {
        GraphHopper hopper = new GraphHopper();
        hopper.setOSMFile(ghLoc);
        // specify where to store graphhopper files
        hopper.setGraphHopperLocation("target/routing-graph-cache");

        // see docs/core/profiles.md to learn more about profiles
        hopper.setProfiles(new Profile("car"));

        // now this can take minutes if it imports or a few seconds for loading of course this is dependent on the area you import
        hopper.importOrLoad();
        return hopper;
    }

    public static void routing(GraphHopper hopper) {
        // simple configuration of the request object
        GHRequest req = new GHRequest(52.54824850398229, 13.385630954186816, 52.51316306397109, 13.391887774811742).
                // note that we have to specify which profile we are using even when there is only one like here
                        setProfile("car").
                // define the language for the turn instructions
                        setLocale(Locale.US);
        GHResponse rsp = hopper.route(req);

        // ...
    }

So I followed route() method, which uses createRouter(), which uses createWeightingFactory(). This method uses DefaultWeightingFactory. If I understand correctly, this can only create successfully a weighting if Profile weighting has the same name as customWeighting which is “custom”:

///....
String weightingStr = toLowerCase(profile.getWeighting());
///.....
        Weighting weighting = null;
        if (CustomWeighting.NAME.equalsIgnoreCase(weightingStr)) {
        //....
         weighting = CustomModelParser.createWeighting(encodingManager, turnCostProvider, mergedCustomModel);

And here I am really lost. I got a couple of questions:

  • First of all, is my understanding correct?
  • Does this mean that customWeighting class is always the one providing/dictating how edge weight are calculated?
  • As you said in your reply:

You can modify the code in CustomWeighting.calcEdgeWeight to do that and inject the new class with your own WeightingFactory

How can I do this? Should I create a new class that extends CustomWeighting ( e.g. ModifiedCustomWeighting) and then override calcEdgeWeighting method with a new method where I get to write the cost function that I want?
And if that’s the case, how can I make sure that when I call route(request), it’s my ModifiedCustomWeighting that is used and not the normal CustomWeighting?

I would highly appreciate any help/guidance as I did not find any clear guide on how to do this.
Thank you very much in advance!