Application examples: Ecological sustainability, Manufacturing industry, Logistics
Value creation: Logistics, Service, Production & supply chain
Development stage: Demonstrator
Company size: 1 - 250 employees
Region: Baden Wurttemberg
EIBA reader with graphic Enlarge

© Circular Economy Solutions GmbH

What are the challenges that needed to be solved?

Remanufacturing, i.e. the reprocessing and reuse of used (industrial) products, is an important part of the circular economy. One of the biggest challenges of remanufacturing is to reliably identify industrial products returned from the market and assess their condition. Today, this process is largely manual, making it error-prone and difficult to scale. In the EIBA project, the project partners, led by Circular Economy Solutions GmbH, are developing an AI-based identification and assessment system for used parts.

What were the concrete benefits?

  • Products that cannot be clearly identified at the end of their life cannot be used for a high-quality circular economy which, like remanufacturing, maintains the product benefit over several life cycles.
  • 85% of the raw material and 55% of the energy can be saved through remanufacturing compared to the production of new parts.
  • EIBA's goal is to identify more used parts through digital technology than it is possible in today's manual process and to avoid identification errors as much as possible.

Application highlight

The employees receive support through sensor technology and artificial intelligence (AI) at their workplaces. The project is developing a system that also observes and evaluates the returned, used product. Sensor technology such as depth cameras or a scale identify the used parts and assess their condition. In addition, business contextual data such as historical returns and product purchases are also analyzed and combined with the sensor data to formulate a recommended course of action for the employee. "We first give the AI already existing data," says project manager Markus Wagner of C-ECO. "After that, the database grows continuously in the process and the AI can acquire further knowledge."

Ident Inspection: Problem broken cover Enlarge

© Circular Economy Solutions GmbH

How can the Industry 4.0 approach be described?

Many industrial products are already suitable for a value-preserving circular economy. However, until now there has often been a lack of incentives and the necessary know-how in the companies to actually have the corresponding parts returned and reprocessed. This is where C-ECO's Industry 4.0 business model comes in. It develops services to return the parts to be remanufactured from the market at the end of the utilization phase in a structured manner and to evaluate them. Around three million used parts are returned annually via a network of 22 logistics hubs worldwide. At present, the identification of parts is still mainly done manually. The biggest challenge is to ensure that the same standards and valuation benchmarks are applied globally to the individual parts.

What does a product consist of? What is usable? Which reprocessing strategy is suitable? To answer these questions, products must be clearly identified and evaluated. Experts often have only a few seconds to do this. However, many product models differ only slightly from one another, and soiling and wear make the evaluation even more difficult.

By using sensor technology, business process data, and artificial intelligence it is possible to make this process less error-prone and easier to scale globally. The goal is not to replace employees but to support them. The different competencies of man and machine are to be combined as efficiently as possible with the goal of identifying used products safely and easily and thus facilitating the circularity of products.

What has been achieved?

The project is not yet completed so no final results are available yet. Regarding the use of sensor technology for machine vision, the first interim results are already available. Based on image data of about 1400 different used parts, 85% were identified correctly in performance tests of the AI. By inserting a hierarchical structure of specialized neural networks into the data, this value has even increased to over 90%. These are tests under laboratory conditions whose validation under real conditions is still pending. Nevertheless, these first results are promising.

What measures were used to achieve the solution?

As an interdisciplinary project consortium EIBA consists of several industry and research partners working together on the objectives. The project partner acatech has conducted surveys on market requirements and acceptance among companies in different industries.

Fraunhofer IPK has trained neural networks and special algorithms for machine vision of the used parts. The Institute for Machine Tools and Factory Management (IWF) at TU Berlin is working on formulating the evaluation of the business process data in combination with the sensor data into a consistent recommendation for action for the employee and integrating this efficiently into the work process. Colleagues from TU Berlin's Sustainable Engineering (SEE) department are evaluating the contribution that the project's developments can make to the circular economy and sustainability. C-ECO coordinates the project and also tests and trials the developed solutions in its CoremanNet logistics hubs.

What can others learn from it?

A core requirement for working with artificial intelligence is the availability of large data sets. These are necessary for efficient training of the algorithms and make profitable AI use possible in the first place. If data is not available to the necessary extent and would have to be generated explicitly for AI use then this would generate high expenses before implementation which would quickly make AI applications appear uneconomical. If the use of AI is planned then it is advisable to take a comprehensive inventory of the relevant company data at an early stage and to plan well how existing data gaps can be closed with little effort, if possible during the ongoing process.