AI-Driven Sorting of Non-Ferrous Materials

The non-ferrous stream consists of a wide variety of objects with varying shapes, dimensions, and material properties. In addition, the objects are constantly changing orientation on the conveyor belt. 

In the current situation, manual sorters are used to identify and sort materials. Due to the shortage of qualified personnel and the need for a stable and repeatable process, there is a demand for further automation. 

The technical challenge lies in reliably recognizing and classifying objects within an unstructured and constantly changing material flow. 

RIWO is developing an intelligent sorting system that combines Vision AI, RGB-D/3D cameras, deep learning, and machine learning. The system continuously analyzes the material flow and creates a digital profile of each object. Based on trained AI models, objects are automatically recognized, classified, and assigned to the correct material category. 

The position and classification data are used to control a FANUC robot. This robot accurately removes selected objects from the material flow and sorts them to their desired destination. 

A key advantage of the chosen architecture is that the system can be flexibly trained to handle new product groups. This allows the facility to be adapted to changing material flows and future sorting needs. 

This application provides customers with a robust solution for robotic sorting in the recycling industry. The system helps reduce reliance on manual labor, ensures consistent sorting quality, improves the purity of material streams, and enhances the utilization of valuable raw materials. In addition, it improves workplace safety and creates a scalable platform for future developments in AI sorting and recycling automation.