What happens to old circuit boards?

Every year, millions of circuit boards end up in landfills — but many of the components on them are perfectly reusable. The problem is that sorting those tiny resistors, capacitors, and chips by hand is painfully slow, so most of them just get shredded and lost.

This project builds a machine that can see, identify, and physically pick components off a circuit board — automatically. At its core is a YOLO11 instance-segmentation model trained on over 7,500 images to recognize 20 different component types with pixel-level precision. It doesn't just draw boxes — it traces the exact outline of each capacitor, resistor, and IC, hitting 84.5% box mAP and 84.0% mask mAP at IoU 50.

Getting the AI to see was only half the challenge. The model's pixel coordinates then need to become real-world millimeter positions accurate enough for a vacuum nozzle to pick up a component smaller than a grain of rice. A degree-3 polynomial mapping and two-speed BLTouch probing bridge that gap, and a modified 3D printer acts as the robotic arm to physically sort each component into the right bin.

The full system — vision, coordinate transform, and pick-and-place — runs end to end with no human intervention. It was presented at the Synopsys Science Fair 2026.