Combining crowdsourcing and mapping customer behaviour in last-mile deliveries

In the light of the dramatic rise of online sales, last-mile deliveries (i.e., the delivery of products ordered online to the final customer) have been increasingly gaining the attention of both managers and academics. As a matter of fact, they are very critical in terms of effectiveness (as customers demand fast and accurate deliveries), and efficiency (since they imply very high costs). Henceforth, logistics players operating in the B2C e-commerce environment are striving to find and implement innovative solutions, different from the costly traditional by-van home deliveries. Among the options analysed by scholars so far, two promising ones are crowdsourcing logistics (i.e., outsourcing delivery activities to “common” people) and mapping the behaviour of customers (i.e., analysing the probability distribution of the customer presence at home and accordingly scheduling deliveries to minimise the probability of failed deliveries). In this paper, we introduce and study a combination between the two solutions, proposing a variant of the Vehicle Routing Problem, which considers both the Availability Profiles and Occasional Drivers (VRPAPOD). We model the delivery problem as a mixed-integer program and solve it with a branch-and-price algorithm. To analyse the benefit of the combined use of crowdshipping and customers availability profiles (APs), we conduct several experiments in a real context in the city of Milan, randomly extracting 100 customers in a 16 km2 area. The combined solution is compared with two benchmarking models, namely the traditional home delivery (traditional VRP) and the crowdsourcing logistics option (Vehicle Routing Problem with Occasional Drivers (VRPOD)). Results prove that logistics players can achieve important benefits by relying on the crowd and scheduling deliveries according to clients' APs, which become more significant in case of high drivers availability.