Vision-Integrated Circuit Printing System
Automated Manufacturing | Computer Vision | Embedded Systems
This project developed a fully automated, vision-integrated circuit printing pipeline combining CNC machining, UR5 robot motion, and ink/component deposition. Within this group project, I was responsible for the design and implementation of the computer vision modules, Python–URP–BUMES system integration, and embedding real-time pass/fail logic into the manufacturing workflow. The system operates at Boston University’s ADML and enables autonomous decision-making during fabrication.
Contributions
Computer Vision Design: Developed three modular OpenCV-based inspection tests:
• CAM Test (average pixel intensity)
• Meander Test (edge density via Canny + thresholding)
• Component Test (morphology + contour filtering)
Integration with BUMES: Wrote new callable functions circuitVision_complete() and runCalibration() inside the mesProcess.py controller to trigger Python vision scripts from the BUMES GUI. The pipeline dynamically redirected substrates based on test results.
Python–URP Synchronization: Coordinated UR5 robotic arm actions (pick/place routines) via custom URP scripts (Project1pickX.urp) referenced inside complete_print_vision.py.
Automated Feedback Loop: Real-time image analysis results determined whether to proceed with ink printing or redirect substrates for rejection—minimizing failure propagation.
Robustness Enhancements: Implemented repeated checks and calibration to address lighting variability and avoid false negatives.