Lesson 10: AI-based sorting technology for waste
The AI-based sorting technology for plastic waste is centralized in Table 6.3.
Table 6.3_ AI-based sorting technology for plastic waste 
Exemple: the ReCircE project
The ReCircE project wants to improve plastic waste-sorting using artificial intelligence (AI) and then issuing a Digital Product Passport to build transparency into the recycled materials chain, making it simpler to re-use plastic granulates from complex products like electric kettles and toys [[i]]. The ReCircE project aims to improve the resource efficiency of material cycles by combining a digital product description – the “life cycle record” – with intelligent sorting technologies supported by artificial intelligence (AI).
6‑4. Figure_ The DLCP stores information along the product lifecycle [https://doi.org/10.1016/j.procir.2022.02.021]
Information about the product and product life cycle is stored in the life cycle file. This includes, for example, the materials used in the manufacturing process and their properties. This information is made usable for material recovery – i.e. for sorting, recycling and subsequent reuse [[i]].
Data can be obtained from the life cycle file and used for improved sorting. Product and material data is made available to machine learning processes to enable AI-based sorting decisions. At the same time, data from sorting flows into the life cycle file and represents a further source of information for subsequent recycling processes. The data from the life cycle file as well as with the help of machine learning methods and sensor-assisted sorting can increase the overall material efficiency in the recycling of products and materials. This means that higher proportions of valuable materials can be recovered and processed into higher-quality products made from secondary raw materials.
The The ReCircE project pursues a solution approach on three levels:
- Informational networking along the value chain: The digital life cycle file connects producers with waste disposal companies and enables simple and efficient communication between different actors. The effects and feedback from actors’ interventions are analyzed and predicted by AI. This enables producers to incorporate experiences from recycling into their product development. Disposers and recyclers, in turn, can fine-tune the sorting and recycling process if they know what the product contains.
- Intelligent sorting and recycling of heterogeneous waste streams: A digitalized sensor-based sorting system is used as part of the project to sort the waste. The sorting system has sensors for color and shape recognition, near-infrared sensors and metal detectors that can be combined in any way. The sorting process is iteratively optimized based on given sensor data and the information from the digital life cycle file. For this purpose, AI decision models are created using machine learning methods that allow specific sorting rules to be generated automatically, including background information such as incompatibilities between materials. With the help of these data-based processes, the recycling process can be significantly improved.
- Resource-efficient optimization of material cycles: The information technology integration of the data from the digital life cycle file and intelligent sorting into the assessment methodology for resource efficiency enables optimization across the entire life cycle and all relevant natural resources. In the form of a tool, different variants of product design, value chains and recycling are optimized in an integrative manner in the sense of sustainable ecological and economic control of material cycles. This should, for example, make it possible to estimate in advance which recycling, i.e. the quality level of the sorting compared to the economic and ecological effort, is suitable.
[[i]] Christiane Plociennik, et al (2022). Towards a Digital Lifecycle Passport for the Circular Economy. Procedia CIRP, Volume 105, Pages 122-127, ISSN 2212-8271, https://doi.org/10.1016/j.procir.2022.02.021