Reconfiguration of manufacturing systems is a frequent and labor-intensive activity driven by evolving products, volumes, and operational constraints. While existing methods based on optimization, digital twins, and expert modeling provide rigorous analysis, they require explicitly specified models and substantial manual effort. This paper presents a multi-agent AI framework for generating geometry-feasible reconfigurable system layouts from updated production requirements. Leveraging 3D machine models, multiple large language model–based agents collaboratively generate and refine shopfloor layout candidates with Unity3D visualization. These high-quality solutions can be optionally optimized further using established reconfiguration methods.
Wang et al. (Fri,) studied this question.