Optimizing e-commerce last-mile vehicle routing and scheduling under uncertain customer presence

Abstract

The recent increase in online orders in e-commerce leads to logistical challenges such as low hit rates (proportion of successful deliveries). We consider last-mile vehicle routing and scheduling problems in which customer presence probability data are taken into account. The aim is to reduce the expected cost resulting from low hit rates by considering both routing and scheduling decisions simultaneously in the planning phase. We model the problem and solve it by the means of an adaptive large neighborhood search metaheuristic which iterates between the routing and scheduling components of the problem. Computational experiments indicate that using customer-related presence data significantly can yield savings as large as 40% in system-wide costs compared with those of traditional vehicle routing solutions.