A production-oriented quality system that predicts a product’s weight before it is physically weighed, replacing a slow manual check with an instant, software-based one.

I built the full signal-to-insight chain. At the front end, the system interfaces with external ADC chips to capture sensor signals and runs real-time processing on them. Those processed features feed an MLP model that predicts product weight to within 0.2% of the true value.

Results surface on a PyQt5 dashboard for on-site operators, while every processed record is synchronized to an SQL database so measurements stay traceable, the foundation for a smart-manufacturing workflow.

This project took first place at the NTNU Mechatronic Capstone Competition (2023) and later became the basis of a peer-reviewed conference paper at the 2023 ARIS International Conference.