In contrast to automated storage and retrieval systems (AS/RSs)
used in warehousing, the cross-docking technology transships
products from incoming vehicles directly to outgoing vehicles by
using the warehouse as a temporary buffer instead of a place for
storage and retrieval, eliminating the need for storage and orderpicking.
However, classic cross-docking models treat the vehicle
routing and scheduling as two separate problems, overlooking the
advantages that can be obtained by treating them as an integrated
one. This paper proposes a bi-objective mathematical formulation
for cross-docking where vehicle routing and scheduling are
cooperatively planned to achieve maximum throughput. A cooperative
coevolution approach consisting of Hyper-heuristics and
Hybrid-heuristics has been developed to deal with the two correlated
problems and achieve continuous improvement in alternating
objectives.
We have experimented our model and algorithms on a dataset
containing four problem instances with real geographical data and
various variable settings. The advantages of the proposed model
and algorithms are shown by an illustrative example and comparative
performances with other existing models and algorithms.
The experimental results show that our cooperative coevolutionary
approach saves around 34% and 10% in transportation cost
and makespan compared to Liao's tabu search with the FCFS
scheduling policy, which treat the vehicle routing and scheduling
separately. The robustness of our cooperative coevolution
approach is verified with statistical tests including the convergence
95% confidence analysis and the worst-case analysis.
In this paper we used the weighting scheme to formulate the
bi-objective model and the weight settings in our experiments are
used for normalizing the two objective values to comparable
ranges. Another approach for tackling the bi-objective problem is
to locate the Pareto front which contains all the non-dominated
solutions. Our future work will be focused on using multiobjective
metaheuristics, such as NSGA-II or MOEA/D, in order to
find the tradeoff Pareto-optimal solution