Crowdsourced last-mile delivery with parcel lockers

Abstract

Crowdshipping is increasingly known as a sustainable solution to address the challenges of last mile delivery (LMD) in urban areas. While employing the crowd to perform LMD appears to be an operationally and financially appealing model, it comes with several challenges in practice, including low willingness to participate in delivery work due to low financial incentive and additional travel effort. Inspired by the Physical Internet concept, in this paper we propose a novel crowdsourced LMD problem and solution approach, which allows a delivery task to be performed by one or multiple crowdshippers using parcel lockers as exchange points. The utilisation of parcel lockers in a crowdshipping network allows for shorter trip detour and better geographical coverage. To achieve this objective, we develop a novel model for locating parcel lockers and allocating delivery tasks. A two-phase algorithm is then developed to rank and choose parcel lockers from the potential locations, which first classifies jobs into single and joint delivery sets and then scores each prospective locker by its utilisation in cooperative delivery. A second algorithm is then designed with three selection strategies of random, roulette, and inverse roulette to assign jobs to crowdshippers for single or joint delivery. To evaluate the performance of the algorithms, experiments were conducted in small and large instances based on a real-world case study. While the exact solution was only capable to deal with small-sized problems, the proposed algorithms were able to produce (sub-)optimal results with significantly low computational expenses. Numerical analyses conducted on large instances showed that enabling joint delivery can improve the success delivery rate by up to 5%, which can be achieved by having a small number of parcel lockers hired at ‘critical’ locations.