I'm trying to use JSprit to solve VRP problem with
FleetSize.FINITE on a predefined set of Services and Vehicles.
The algorithm is quite fast to execute, however the result is not good from a geographical point of view.
What I mean is that, after plotting result to a map, I see that routes for Vehicles frequently cross themselves.
If one looks at the results I'm getting may think that there is no or few logic behind the algorithm.
How can I achieve something that is clustered in a better way?
Also, there's a way to get almost the same result (I know exactly the same is not possible, but at least comparable) when starting from the same set of Vehicles and Services?
I'm newbie of JSprit and I'm using the SimpleExample case to test the Service allocation on Vehicles.