Collaborative Research: SmartSOM: Self-Organizing Manufacturing with Task Constraints and Social-Organizational Learning

Personalized production is transforming manufacturing by enabling products to be tailored to individual customer needs. While this trend creates new opportunities for innovation and economic growth, it also introduces unprecedented complexity and volatility into manufacturing operations. Traditional manufacturing systems rely heavily on centralized planning and control, which becomes increasingly difficult to maintain as product variety, order fluctuations, and process interdependencies grow. Future manufacturing systems must therefore be capable of continuously adapting to changing conditions while maintaining efficiency, quality, and reliability. This project addresses this challenge by advancing the scientific foundations of self-organizing manufacturing systems that can coordinate production activities through distributed decision-making rather than centralized control. The outcomes are expected to contribute to more resilient and responsive manufacturing enterprises, strengthen the competitiveness of domestic manufacturing, and support workforce development by engaging graduate and undergraduate students from two universities. Educational materials, simulation tools, and research findings will be disseminated broadly to support learning and future innovation in intelligent manufacturing. This project develops a next-generation self-organizing manufacturing systems integration framework for personalized and dynamic production. The research investigates how autonomous manufacturing agents can collectively perform dynamic scheduling, resource allocation, and process control under rapidly changing production conditions. Three complementary mechanisms are integrated into the framework. First, task constraints are used to identify and prune infeasible actions, enabling agents to focus on effective decisions while reducing learning complexity. Second, observational social learning allows agents to improve their behavior by learning from the states and experiences of other agents. Third, organizational learning enables the formation of adaptive behavioral norms that regulate the balance between decentralized autonomy and coordinated action. The research employs multi-agent reinforcement learning, large-scale simulation experiments, and quantitative measures of manufacturing performance, robustness, adaptability, and learning efficiency. The resulting framework is expected to advance understanding of how distributed intelligent systems can achieve efficient self-organization in complex manufacturing environments and establish new principles for the design of future manufacturing systems capable of responding to uncertainty, personalization, and rapid change. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.