Abstract
We study a variant of the Probabilistic Travelling Salesman Problem arising when retailers crowdsource last-mile deliveries to their own customers, who can refuse or accept in exchange for a reward. A planner must identify which deliveries to offer, knowing that all deliveries need fulfilment, either via crowdsourcing or using the retailer's own vehicle. We formalise the problem and position it in both the literature about crowdsourcing and among routing problems in which not all customers need a visit. We show that to evaluate the objective function of this stochastic problem for even one solution, one needs to solve an exponential number of Travelling Salesman Problems. To address this complexity, we propose Machine Learning and Monte Carlo simulation methods to approximate the objective function, and both a branch-and-bound algorithm and heuristics to reduce the number of evaluations. We show that these approaches work well on small size instances and derive managerial insights on the economic and environmental benefits of crowdsourcing to customers.

Abstract
We propose a new approach for the lastmile delivery problem where, besides serving as collecting points of orders for customers, parcel lockers are also used as transshipment nodes in a 2-echelon delivery system. Moreover, we consider that a customer (occasional courier) visiting a locker may accept a compensation to make a delivery to another customer on their regular traveling path. The proposed shared use of the locker facilities - by customers that prefer to self-pick up their orders, and also as a transfer deposit for customers that prefer home delivery - will contribute to better usage of an already available storage capacity. Furthermore, the use of occasional couriers (OCs) brings an extra layer of flexibility to the delivery process and may positively contribute to achieving some environmental goals: although non-consolidation of deliveries may, at first sight, seem negative, by only considering OCs that would go to the locker independently of making or not a delivery on their way home, and their selection being constrained by a maximum detour, the carbon footprint can be potentially reduced when compared to that of dedicated vehicles. We present a mixed-integer linear programming formulation for the problem that integrates three delivery options - depot to locker, depot to locker followed by final delivery by a professional fleet, and depot to locker followed by final delivery by an OC. Furthermore, to assess the impact of OCs' no show on the delivery process, we extend the formulation to re-schedule the delivery of previous undelivered parcels, and analyze the impact of different no-show rates. Thorough computational experiments show that the use of OCs has a positive impact both on the delivery cost and on the total distance traveled by the dedicated fleets. Experiments also show that the negative impact of no-shows may be reduced by using lockers with higher capacities.

Abstract

Abstract
Scope and Mission

Abstract
Nowadays there are several countries running independent kidney exchange programmes (KEPs). These programmes allow a patient with kidney failure, having a willing healthy but incompatible donor, to receive a transplant from a similar pair where the donor is compatible with him. Since in general larger patient-donor pools allow for more patients to be matched, this prompts independent programmes (agents) to merge their pools and collaborate in order to increase the overall number of transplants. Such collaboration does however raise a problem: how to assign transplants to agents so that there is a balance between the contribution each agent brings to the merged pool and the benefit it gets from the collaboration. In this paper we propose a new Integer Programming model for multi-agent kidney exchange programmes (mKEPs). It considers the possible existence of multiple optimal solutions in each matching period of a KEP and, in consecutive matching periods, selects the optimal solution among the set of alternative ones in such a way that in the long-term the benefit each agent gets from participating in the mKEP is balanced accordingly to a given criterion. This is done by use of a memory mechanism. Extensive computational tests show the benefit of mKEPs, when compared to independent KEPs, in terms of potential increase in the number of transplants. Furthermore, they show that, under different policies, the number of additional transplants each agent receives can vary significantly. More importantly, results show that the proposed methodology consistently obtains more stable results than methodologies that do not use memory. © 2020 Elsevier Ltd