Generative Model Predictive Control in Manufacturing Processes: A Review

This review has demonstrated the potential of generative ML technologies to enhance MPC for manufacturing processes. We surveyed key generative ML methods and their integration with MPC, emphasizing their unique ability to capture uncertainty and provide high-fidelity surrogates that align with advanced manufacturing requirements. In doing so, we identified critical gaps that still prevent reliable deployment, motivating the need for new research directions. Taken together, the discussion highlights that generative ML is not simply an incremental add-on to MPC but a means to reshape control for manufacturing. By enabling data-driven representations of uncertainty, producing more reliable state information, and supporting control strategies in MPC, generative ML-driven MPC moves manufacturing systems closer to robust and flexible autonomy. The multi-robot example illustrates how these benefits can materialize in practice; however, the implications extend broadly to other manufacturing domains facing similar challenges. Ultimately, the integration of generative ML with MPC offers a pathway toward next-generation manufacturing control systems that are resilient to uncertainty, adaptable across tasks, and capable of sustaining performance under real-world complexity.