Physics-Based Sensing and Machine Learning for Smart Manufacturing

As the fundamental building blocks of Industry 4.0, sensing and artificial intelligence play a critical role in advancing the state-of-manufacturing. The ability in acquiring data in-situ and extracting clues from the data to guide the action of assistive infrastructure such as robots is essential to enhancing process control and production planning.

This seminar highlights research on manufacturing process-embedded sensing and ma-chine learning for smart manufacturing, illustrated in two examples. The first example de-scribes the design and experimental evaluation of a multi-physics sensor with acoustic-based wireless data transmission capability for the online quantification of melt temperature, pressure, velocity, and viscosity within an injection mold. The second example illustrates machine learn-ing methods for the recognition of current actions and prediction of future actions of human operators during assembly operations, which provide the prerequisites for human-robot collab-orative assembly. The presentation highlights the potential of use-inspired basic research in advancing the state of manufacturing.