Strengthening the circularity of products and components is necessary in order to conserve resources and reduce the amount of waste. Remanufacturing, which in combination with subsequent reuse extends the service life of products, offers one way of doing this. This is why the German Circular Economy Act gives remanufacturing a high priority, as it offers ecological and economic potential. For example, a life cycle assessment of old car parts calculates that remanufacturing an exhaust gas turbocharger can reduce_{CO2 equivalent} emissions by 37 percent compared to new production. In order to increase the aforementioned potential for sustainability, reverse logistics is required in which the old parts are collected and sorted. This process is known as sorting and involves identifying and assessing the condition of old parts. A major challenge here is the high product diversity, which requires individual and manual handling.

At the same time, no employee can distinguish between all variants. The only reliable identification feature for them is the labeling with part numbers. However, used products are often dirty, deformed and no longer clearly identifiable. Old parts are then often incorrectly sorted out.

This offers potential to make reading more sustainable through technical innovation. This is what the EIBA research project is investigating: its aim is to develop an AI-supported system for semi-automated reading.

The framework for the development is the sorting process that Circular Economy Solutions GmbH already carries out for the reverse logistics of automotive components. For partial automation, Fraunhofer IPK is developing a digitization system and AI-based object recognition for reading based on 2D and 3D image data and weights. In addition, the HAMSTER department at TU Berlin is evaluating the business and process data in order to support the employee with integrated human-machine interaction and generate process stability. At the same time, the SEE at TU Berlin collects data and information for a sustainability assessment. This is used to check whether environmental and social impacts are reduced as a result of the process change.

The German Academy of Science and Engineering (acatech) is interviewing suitable companies from the reverse logistics sector in order to understand the requirements and demand for AI-based solutions from companies in this sector. In this way, the project results can be used to determine which companies can benefit from the developed system.

Within the project, the available knowledge about a component in the form of data is combined with human experience and stored in a holistic knowledge management system. Through the continuous digitization and use of data, knowledge about each product variant is constantly expanded, analyzed and evaluated and corrected through interaction with the user. The combination of image, component and business data is a new, innovative and holistic approach from current research into decision-making. Combining human skills with AI-based analysis results in robust and objective decisions based on the "four-eyes principle". At the same time, these are documented with relevant data and are therefore traceable in retrospect.

Reading of generators a) Identification by labeling two variants with amperes. b) Condition assessment of the same generator c) Poorly recognizable component numbers of different old parts. © EIBA

The "four-eyes principle" of employee and AI leads to greater transparency and efficiency in the process and reduces the workload for the employee.

As part of the AI-based analysis, image-based object recognition enables identification independent of part numbers. Initial results already show a recognition accuracy of around 96 percent. With a realistic quantity of one million old parts per year, 5 to 7 percent, i.e. up to 70,000 of them, are sorted out. With the help of AI-based object recognition, it is expected that 67,200 (96%) more old parts can be correctly returned to the cycle than before.

Justus Caspers, TU SEE; Hannah Lickert, TU HAMSTER; Clemens Briese and Marian Schlüter, Fraunhofer IPK, Machine Vision

EIBA, www.linkedin.com/company/eibaprojekt

Briefly explained:

Fraunhofer Institute for Production Systems and Design Technology IPK

As a production technology research and development partner with strong IT expertise, the Fraunhofer IPK in Berlin offers system solutions, individual technologies and services for digitally integrated production.

TU Berlin Institute for Machine Tools and Factory Management (IWF) - Department of Handling and Assembly Technology. (HAMSTER)

The department's expertise includes the design, experimental and simulative testing of processes, the digital control and optimization of processes and the investigation of coupled process influences for the analysis of interdependencies and the optimization of process boundaries.

TU Berlin - Department of Sustainable Engineering (FG SEE)

The department develops scientific solutions for the sustainability assessment of products and technologies.

Briefly explained: The MHI e.V.

The Wissenschaftliche Gesellschaft für Montage, Handhabung und Industrierobotik e.V. (MHI e.V.) is a network of renowned university professors - institute directors and chair holders - from German-speaking countries. The members conduct both fundamental and application-oriented research on a wide range of current topics in the fields of assembly, handling and industrial robotics. Further information on the society, its members and activities: www.wgmhi.de