I'm reading some papers about Delivery Clustering to solve following generic problems:
Given N orders (with its Pick-up and Drop location point) and M delivery-man (with his location). We are going to assign M' drivers (M' <= M) to handle that N orders. Each driver can handle some X orders (X >= 0, in case X = 0, driver is not assigned). Our goal is to minimize total driving cost of drivers, including driving time from his location to the first pick-up order's location, and whole driving time to delivery all X orders.
In mainly papers I've noticed that the authors just considered these N orders as one point (not separated as 2 pick-drop points) and for each cluster, they will find the best assignment to minimize total driving cost from current location to the first order in that cluster.
How can we treat two pick-drop points of order within same cluster?