<p>To improve delivery efficiency and reduce delivery costs in last-mile delivery, this study explores a vehicle routing problem in which the courier provides simultaneous pickup and delivery services to customers with roaming locations. The problem is formulated as a mixed integer linear programming model to minimize the total travel cost. We then develop a two-stage metaheuristic that combines a random selection greedy insertion algorithm and a large neighborhood search algorithm. The computational results show that our algorithm has certain advantages in solution quality and computational time compared to the commercial optimization solver. Moreover, incorporating simultaneous pickup and delivery into the vehicle routing problem with roaming delivery locations can significantly reduce the total delivery cost as the number of customers increases.</p>