Multi-objective parallel robotic dispensing planogram optimisation using association rule mining and evolutionary algorithms

Abstract

This research addresses a medication planogram optimisation problem for robotic dispensing systems (RDSs) in mail-order pharmacy automation (MOPA) facilities. A MOPA is used by a high-throughput fulfilment facility that processes a large volume of prescription orders. In MOPA facilities, each RDS unit integrates auto-dispenser devices and a robot arm to count and dispense medications automatically to complete high demand. An RDS planogram is the allocation of medications in one RDS unit and their distribution in different RDS units. A significant challenge in MOPA systems is to design an efficient planogram strategy. In this study, the RDS planogram is optimised to meet three objectives: association between medications, workload balance of RDSs, and robot arm travel distance. Association rule mining (ARM) is applied to explore the associations between medications, whereas a nonlinear mixed-integer programming (MIP) model is developed to optimise medication allocation based on ARM outputs. Four evolutionary algorithms, namely Non-dominated Sorting Genetic Algorithm (NSGA-II), knee-based NSGA-II (k-NSGA-II), Pareto Archived Evolution Strategy (PAES), and Strength Pareto Evolutionary Algorithm (SPEA-II), are applied to solve the proposed planogram optimisation model on eight experimental problems. Based on the different performance evaluation criteria, the best algorithm with higher performance is identified for each criterion.