Last Mile Delivery Considering Time-Dependent Locations

ABSTRACTLast-mile delivery (LMD) is known as the task of delivering items from a warehouse to a customer. In LMD's original version, each item is to be delivered to a single location in a city's network. In this paper, we extend such a scenario by assuming that there can be alternative delivery locations for each item, depending on the time of the day, and aim at maximizing the number of deliveries completed by a courier. We call this new problem the Last Mile Flexible Delivery (LMFD) problem. In fact, at least one e-commerce company, namely Amazon, currently offers the possibility of delivering items to a customer's parked car, thus making the LMFD problem, not only computationally challenging, but also practically relevant. After showing LMFD's NP-hardness, we propose an efficient heuristic approach based on provably correct pruning of the search space, and, using a real and large mobility dataset, we show that such heuristic is both effective and highly efficient.