Retrieval request sequencing in a novel tier-captive AVS/RS with a double-capacity lift

We study an innovative autonomous vehicle storage and retrieval system equipped with a double-capacity lift (AVS/RS-DL), where the lift is capable of transporting two unit loads simultaneously. Therefore, given a set of retrieval requests, the retrieval sequencing problem consists of deciding which requests are paired together in the same tour and determining the sequence in which the requests are processed to minimize the makespan. The problem is proved to be NP-hard. We propose a two-step decomposition algorithm that first employs a greedy heuristic to identify pairs of retrieval requests. Subsequently, a dynamic programming approach, complemented by a rollout heuristic, is utilized to determine the sequence of these pairs. Comparative numerical experiments demonstrate that the algorithm leads to substantial makespan reduction compared to the methods proposed in the literature. Furthermore, we examine the benefits of the double-capacity lift against single-capacity alternatives and separate lifting tables that can operate in parallel but not pass each other. Real-world data analysis showcases the superiority of our proposed algorithm over the commonly employed first-come-first-served policy. Finally, we demonstrate that our algorithm can easily be extended to accommodate varying buffer sizes, a multi-capacity lift, and to simultaneously sequence both retrieval and storage requests.