A Non-Sorted Metaheuristic Method for the Multi-Objective Job-Flow-Shop Scheduling Problem in Conflict-Free Robot Swarm Manufacturing

Robot swarm manufacturing is a promising direction in smart manufacturing that aggre-gates multiple robots to collaboratively complete production jobs; however, achieving conflict-free scheduling remains a significant challenge. Traditional methods struggle to address this issue since robot swarms are inherently prone to conflicts. This article puts forward a non-sorted metaheuristic method to solve it. First, the conflict-free robot swarm manufacturing problem—integrating a multi-objective optimization problem (MOP), a flexible job-shop scheduling problem (FJSP) for job processing, and a flow-shop schedul-ing problem (FSP) for robot travel—is formulated as a multi-objective job-flow-shop scheduling problem (MJFSP). The robot swarm must accomplish all manufacturing jobs while achieving high manufacturing performance, energy efficiency, and conflict-free op-erations. Second, a non-sorted metaheuristic algorithm based on an artificial plant com-munity (APC) is proposed. It employs a sequential-pairwise single-elimination tourna-ment system (SSTS) to select elites with a time complexity of 𝑂(𝑛), which scales linearly with the population size (𝑛). This surpasses the sorting-based elite selection with poly-nomial time complexity employed in most metaheuristic methods, such as the 𝑂(𝑛²) of the non-dominated sorting genetic algorithm-III (NSGA-III). Third, an MJFSP benchmark dataset is built, and the experimental results uncover the complex dependencies between the FJSP for job processing and the FSP for robot traveling. The proposed method im-proves the makespan by up to 13.10% and reduces non-loaded energy consumption by up to 13.49%, achieving zero collision time and an average solution time 11.18% faster than NSGA-III.