A Two-stage Stochastic Programming for AGV scheduling with random tasks and battery swapping in automated container terminals

Abstract

Automated guided vehicle (AGV) is one of the main equipment for horizontal transportation in automated container terminals, and the optimization of AGV scheduling has become increasingly important. Existing scheduling systems tend to make decisions based on deterministic conditions, ignoring the dynamic changes and uncertainties of the terminal environment, such as the arrival of random tasks during AGV operations. In addition, the battery swapping process is neglected in most AGV scheduling studies, yet it is crucial to ensure the operation of AGVs. In this paper, we construct a two-stage stochastic programming model for the joint scheduling problem of battery swapping and task operation with random tasks. A double-threshold constraint for battery swapping decision-making is adopted. The results show that the double-threshold strategy is better for AGV utilization than the single-threshold one. Upon the solution method, a simulation-based ant colony optimization algorithm is proposed. Sample average approximation is used to calculate the expected cost, and two local search procedures are introduced to improve the quality of the solutions. In the cases of multiple instances and several random task samples with different arrival rates, our method was compared with three practical policies under a deterministic model. Computational results show that the scheduling scheme considering random tasks in advance is more robust and stable.