The EIBA project focused on developing a sensor‑based, machine‑learning system for the automated identification and assessment of used industrial components. Over 600 000 assessment operations were recorded and continuously analysed to train and validate computer‑vision and data‑analysis models. The team employed deep‑learning techniques, refined existing algorithms, and created new application‑specific methods. Fusion algorithms weighted the AI outputs, producing concise identification suggestions that were integrated into the production workflow and the human‑machine interface. Employee surveys assessed the acceptance and perception of the EIBA system.

In a production‑environment validation, the system achieved a 15 % higher identification accuracy compared with the current state of the art (SdT). Specific gains were 9 % for diesel injectors and pumps, 19 % for starters, and 14 % for alternators. The project’s research indicates that the developed AI models could raise accuracy by up to 28 % relative to industrial SdT systems. During the validation phase, the overall process time with EIBA was 13.5 % longer than the manual SdT, mainly due to camera‑capture wait times and the comparison against highly experienced operators who perform the manual process daily. The team expects that targeted software and hardware improvements after the project will eliminate this speed disadvantage and may even provide additional throughput benefits without compromising quality.

From an ecological perspective, the energy required for operating and training the AI was shown to be offset after identifying 15 additional parts that can be remanufactured, using starters as a case study. If the EIBA technology were deployed globally by C‑ECO, the annual CO₂‑equivalent savings for this product group alone would reach approximately 27.5 t. The project also produced a dataset comprising roughly 500 distinct industrial parts and the corresponding trained machine‑learning models, which will be made available for further research. Numerous scientific publications have already been released by the partners, underscoring the importance of high‑quality data for successful digitalisation of business processes and the challenge of fostering a culture that demands data quality alongside product quality.

Collaboration was organised under the coordination of Circular Economy Solutions GmbH, with Fraunhofer IPK, the Technical University of Berlin (including the Institute for Machine Tools and Factory Operation and the Department of Handling and Assembly Technology), acatech, and TU Berlin SEE as consortium partners. The project ran for 45 months from 1 September 2019 to 31 May 2023 and was funded by the German Federal Ministry of Education and Research under grant number 033R226A. C‑ECO plans to further industrialise the EIBA technology and integrate it into its CoremanNet service, enabling the scalable transfer of specialist knowledge and quality improvements across its network of reading stations. The consortium’s experience demonstrates that the EIBA approach can be extended beyond automotive to other sectors such as machinery and consumer goods, thereby broadening the impact of circular‑economy digital solutions.