Research Interests

1) developing cost-effective edge devices together with data management, processing, and learning pipelines to reduce the barriers for manufacturers in building affordable digitalized and intelligent shop floors. 2) improving AI and ML credibility and generalizability in manufacturing, through integrating domain-knowledge-based constraints to data-driven learning processes and, toward mutual augmentation between data-driven decision-making and physical understandings. 3) reducing demands/efforts on manual data labeling and ML model maintenance, through unsupervised learning from big unlabeled plant data on characterizing machine/process dynamics/uncertainties. 4) advancing in-situ process monitoring (i.e., defect detection and quality prediction) and real-time control of complex manufacturing processes (e.g., metal additive manufacturing). 5) robotic automation of complex manufacturing processes (e.g., welding).