Automatic Recognition and Extraction of Materials


The present project is aimed at developing a system capable of performing automatic recognition of objects made of different materials and separating them, always in a completely automated manner.

The automatic material recognition system is based on artificial intelligence techniques, a new technological frontier that today’s state of the art has led to practical applicability; the extraction of the selected objects is based on advanced robotics and mechatronics.

So far, artificial intelligence techniques have reached an adequate maturity level, and our company has also developed skills in this sector.

The selection of objects in the AREM system exploits the reflection and absorption characteristics of different materials at various wavelengths of the electromagnetic spectrum, in the fields of visible light and infrared.

For this purpose, an infrared video camera (NIR) and an RGB stereo video camera are used to recognize all types of plastics, with a success rate close to 100% [1 PET, 2 HDPE, 3 PVC, 4 LDPE, 5 PP ( polypropylene), 6 PS (polystyrene), 7 ABS], metals, wood and paper.

This project has, as its purpose, the possible applicability in the separation of urban solid waste or industrial waste, a field and a problem that surely needs technologically advanced solutions.

This system uses a hyperspectral sensor (RedEye 1.7), an RGB-D video camera (Intel Realsense D435), a robotic arm with 6 degrees of freedom (Kuka KR 6 Agilus), a suction gripping device, a special illuminator and a conveyor belt. The software is developed in C++ programming language, using the ROS framework and the OpenCV vision library.

The Engineering Department of the Roma Tre University (AREM) and the Stam S.r.l. company cooperate with IES.

This project is developed thanks to the financial support of the FESR funds.