In these two collaborative projects funded by NCCR Automation, we aim to develop methods for advancing manufacturing systems' productivity, resilience, and efficiency within the framework of Industry 4.0 by focusing on two topics:

1. Control and Task Planning for Advanced Manufacturing using expert feedback and domain knowledge: This project focuses on improving control and task planning in advanced manufacturing by utilizing domain knowledge and expert feedback. We incorporate methods from domain adaptation, transfer learning, and large-language models to refine process planning. Key objectives include:
2. Lifelong Learning and Adaptation Using Digital Twins: This project aims to use digital twins for the continuous optimization of industrial processes. Techniques such as reinforcement learning, continual learning, Bayesian optimization, and adaptive control will be employed to facilitate continuous learning and adaptation.

The major goals are:

• Establishing dynamic digital twin models for real-time monitoring and optimization of manufacturing processes.
• Implementing reinforcement learning strategies to enable adaptive control over time.
• Utilizing Bayesian optimization to fine-tune process parameters continually, ensuring optimal performance and productivity.

These methods increase flexibility in manufacturing, and enable individualized production, collaborative manufacturing, and zero defect manufacturing.