The disposal of plastic waste is one of the important environmental issues of our time: mountains of plastic pile up in emerging and developing countries, microplastics pollute the world's oceans. A group of scientists from TU Darmstadt, Fraunhofer IWKS and the German Research Center for Artificial Intelligence (DFKI) is using Artificial Intelligence to develop a new recycling process to solve this global environmental problem.

Near-infrared spectrometry is a proven sorting method that has been used in large recycling halls for years. It makes it possible to separate plastic products such as films, bags and packaging from the rest of the waste. The technology reliably recognizes the most common polymers and can classify them automatically. However, this does not work for plastics from complex products with a very high diversity of materials, which may also contain pollutants. Examples of this are electronic or clothing articles.

Theoretically, molecular chains of plastics can be reused up to 20 times. However, the recycling process would have to be significantly improved for this to happen. This is precisely where the joint project "Digital Lifecycle Record for the Circular Economy", or ReCircE for short, comes in. "Our goal is to combine sensor-based sorting processes with the possibilities of machine learning," says project leader Christiane Plociennik, a staff member at the DFKI's Innovative Factory Systems research unit.

Better analysis of complex plastic compounds

The scientists are currently training an AI system with data from near-infrared sensors, metal detectors and 3D object construction to better analyze these complex plastic compounds. In the end, the system should detect individual molecular chains in the plastic that is later melted down at an early stage and process it in such a way that it breaks down into the respective fractions - i.e. the chemical subproducts.

The goal is to obtain four or five pure grades whose quality is comparable to the original plastic. As a partner in the project, Papier-Mettler, a European market leader for service packaging, will test the new plastic in practice.

Full transparency in the recyclables chain

As part of ReCircE, the project partners are also working on the creation of a so-called digital product passport. This is intended to ensure transparency throughout the entire recycling chain in order to facilitate the recycling of plastics from highly developed products. An AI system scans the data from the product passport and sorts the plastic accordingly during waste separation.

ReCircE was launched by the Rhineland-Palatinate technology initiative SmartFactory KL, in which the German Research Center for Artificial Intelligence (DFKI), TU Darmstadt, Fraunhofer IWKS and the companies Green Delta and Papier-Mettler work together.