An effective metaheuristic for the last mile delivery with roaming delivery locations and stochastic travel times

Abstract

We consider the last mile delivery system with roaming delivery locations and stochastic travel times. The problem is formulated as a two-stage stochastic programming model with recourses, its objective is to minimize the total travel time of serving a set of customers with roaming delivery locations under stochastic travel times. We propose an effective metaheuristic integrating a sampling strategy (sample average approximation, SAA) to solve stochastic model, our metaheuristic consists of random selection-greedy insertion (RSGI) and hybrid iterated greedy algorithm with route reoptimization and simulated annealing (HIGRR-SA). The computational study shows that our solution method has an advantage in terms of solution quality and computational time. The comparison of the stochastic and deterministic solutions shows that a significant reduction in expected travel time can be achieved as well as a smaller fluctuation considering stochastic information.