An end-to-end automated photoelastic inspection instrument that I led from proof-of-concept to integrated system, and that is now under patent review (Taiwan/US applications filed). It combines non-destructive optical inspection with AI to make quality assessment faster, more repeatable, and ready to slot into an automated line.

The work spanned every layer of the stack:

• Mechanism & optics: a modular device with adjustable optics for high-precision camera positioning, iterated through multiple rounds of SOLIDWORKS design and 3D printing with maintainability in mind.
• Firmware & control: non-blocking microcontroller firmware driving a stable, repeatable automated inspection workflow, with both manual and fully automated modes.
• AI pipeline: a hybrid Autoencoder / MLP model with OpenCV that infers residual stress from the optical signal, running fast enough on edge hardware for in-line use.
• Interface: a PyQt5 GUI to visualize inspection results and tune hardware parameters in real time.

The prototype was invited to Intelligent Asia 2025 at the Taipei Nangang Exhibition Center, where we presented it to both industry and academia. The project was developed in the Polymer and Smart Strategic Processing Laboratory (P2SPLAB) under the guidance of Prof. Kun-Cheng Ke.