A Flexible Approach to PCB Characterization for Recycling

The rapid growth of electronic waste (e-waste) highlights the need for effective recycling processes. Printed circuit boards (PCBs) are a significant component of e-waste, containing valuable materials and toxic elements. However, the recycling of PCBs faces challenges associated with their diverse materials and components, lack of standardization, and high costs. Current practice involves manual sorting, which is suboptimal, and automation is necessary. This article proposes a novel solution to PCB characterization for recycling, using a simple RGB camera to locate and classify three types of PCBs on a conveyor belt. The approach consists of a modular architecture that combines deep-learning solutions to segment PCBs, identify single components, and classify them. The architecture design considers the requirements of a robotic solution for sorting PCBs, and it has been tested in challenging scenarios.