Community logistics and dynamic community partitioning: A new approach for solving e-commerce last mile delivery

Last mile delivery shows an increasingly tough challenge for logistics service providers due to the rapidly expanding e-commerce sales around the globe. To ease the implementation of last mile delivery, an effective delivery strategy is to predetermine the service regions of vehicles before optimizing their delivery routes. On this ground, this paper proposes a new delivery strategy named Community Logistics (CL) to generate vehicle service region and departure time dynamically. Through adopting this new delivery strategy, we transform the original last mile delivery to a new type of research problem, namely dynamic community partitioning problem (DCPP), with an aim to strike a balance between vehicle service region range, order delay time and vehicle capacity usage based on the real-time order arrivals and vehicle availability status. We present a Markov decision process (MDP) model for the DCPP and develop a heuristic solution approach to solve this MDP model. Numerical results demonstrate significant benefits of the proposed solution framework and delivery strategy.