Routing for couriers

[kenneth]

Experience shows that conventional route planners are not very attractive for couriers or distributors in general. The routing problem faced daily by such users involves variable
customer locations, variable drop sizes and relative large volumes. What is more, data is not always correct and in many depots, sorting is still done manually. By consequence, the
incomplete data, the large volumes and the costly sorting complicate the routing process in such a way that conventional route planners turn out unfit. The major problem here is the
high degree of flexibility that they require, as their produced routings may change integrally every day.

For that reason, couriers often rely on customer regions and fixed routes. Customer regions for one, decrease the amount of entities planners have to deal with and make the routing process
more transparant. Fixed routes on the other hand subsequently facilitate the organization of routing, sorting, loading and the assignment of drivers.

In our study, we assume such high-level planning with customer regions and fixed routes is given and deal with the problem on operational level, where day-to-day changes in workload
may render the original planning infeasible. We investigate a multi-objective heuristic that is able to recombine customer regions into new routes, taking into account the compactness of
the routes, the resemblance to the baseline solution and the workload balance.