An adaptive ant colony algorithm for crowdsourcing multi-depot vehicle routing problem with time windows

Logistics distribution faces difficulties such as increasing delivery demands, demanding delivery time and lower delivery costs. However, the routing optimization for logistic distribution is not a simple vehicle routing problem (VRP). This study focuses on crowdsourcing multi-depot vehicle routing problem with time windows (MDVRPTW) problem, permitting one crowdsourcing driver to deliver multiple customers, which has not been covered yet. To quickly distribute many delivery tasks within a certain time, this paper proposes a two-stage MDVRPTW framework. The first stage is the division stage in which a k-means algorithm based on the elbow method is used to transform the multi-depot problem into several single-depot problems. Then a crowdsourcing VRPTW model is developed, which is a mixed-integer linear programming model. The second stage is the optimization stage, for which an adaptive ant colony (ADACO) algorithm is proposed to determine the most reasonable distribution routes for each distribution center. Solomons classic VRP data is then used to validate the models’ effectiveness, with the results confirming that the crowdsourcing MDVRPTW strategy could save 10.02% costs than traditional MDVRPTW strategy. By comparison, the ADACO algorithm could significantly improve the ACO’s weakness of falling into a local optimum, and would be more suitable for solving large-scale vehicle routing problems than the GUROBI solver.