Hi, I am solving Vehicle Routing Problem where customer demand is uncertain (random variable with some distribution). Does Jsprit provide a way to tackle this kind of problem ?

jsprit always needs concrete problems (i.e. concrete jobs and vehicles - not random variables). if you can model your problem as sequence of concrete problems (that might represent your distribution), jsprit might help.

Suppose we know how much to pick up from each shop and then I do clustering and find centroids of the clusters. Now if I plan to run VRP on centroids of cluster with vehicle with large capacity (say truck)in such a manner that truck moves through the centroids, and it stops at centroid for some time to pick the goods collected by bikes that are visiting the current cluster.

From VRP point of View, can we transfer volume picked from one vehicle to another when we know the place and TimeWindow at which transfer has to occur. In this way, a truck has bikers ( or van) associated with it.

This can be helpful because now if bike is full, biker does not need to return to depot to unload all goods and then go to pick more goods. Biker only needs to unload it at some place already known to him which is also closer to him.

I am thinking of using “Break” feature wherein bikers takes break at particular place. The problem is, can at this point biker unload its picked volume?

Second way could be: We first run VRP where each pickup point has location that of centroid and volume to picked up is equal to total volume associated with the cluster and for vehicle we consider only Truck. And after running the VRP, we store the path information. We also know what is time window in which the truck is going to be there. Suppose path is indexed from 1 to n. Now I am going to run VRP second time wherein bikers or van associated with truck for a given i ( i = 1,2, … n) visits the ith cluster and then visits (i+1)th cluster and so on. Bikers and van are allowed to unload the goods at centroid which is represented by imaginary truck.