Most plastic waste gets down-cycled into lower-quality products because recycling plants cannot reliably handle the variable mix of degraded and contaminated materials that arrives at their facilities. This project aims to fix that variability problem. Instead of sorting plastics by simple physical properties like density or colour—which often fails—the researchers will build an AI system that learns to predict how a given batch of waste will behave during reprocessing. The system will then decide the best way to upgrade that batch into a product that meets a specific quality specification, such as a new bottle or a food container. If it works, manufacturers could use far more post-consumer recyclate and far less virgin polymer in high-value products. The impact would be felt across supply chains that currently rely on virgin plastic: packaging, automotive parts, construction materials, and consumer goods. The technology would not eliminate plastic use, but it would make the recycling loop genuinely closed, keeping material in circulation rather than sending it to incineration or landfill. The project is applied engineering, not fundamental science—its success depends on whether the algorithms can handle real-world waste streams reliably.
View original technical description
The circular economy, aiming for zero-waste in plastics, necessitates a multifaceted approach: (i) eliminating the unnecessary use of plastics, (ii) developing innovative designs that are easily recyclable or reusable, and (iii) reintroducing larger quantities of post-consumer (PCR) or post-industrial (PIR) recyclate into high-value products — the latter being the focal point of the PLASTIC proposal. One of the primary reasons plastics currently undergo down-cycling rather than true recycling is the significant variability in the quality of waste streams. This variability arises from differences in degradation levels and contamination though e.g. mixed waste plastics. To establish a circular economy based on closed-loop recycling, where products can be recycled back into the same product or products of similar quality, we must advance intelligent sorting, recycling, and remanufacturing processes. These processes should effectively eliminate the prevailing fluctuations in quality and composition. In the PLASTIC proposal, we intend to leverage artificial intelligence (AI) and machine learning (ML) principles to develop intelligent plastic sorting and mechanical recycling systems. These systems will employ computer algorithms that continually enhance their performance through experience. The developed system will possess the capability to predict the processability and properties of plastic waste with variable quality. It will then utilise this information to make informed decisions regarding the most efficient upgrading and remanufacturing procedures for a given product specification. Our ultimate objective is to maximise the PCR or PIR content in recycled products and minimise the use of virgin polymer in end-products moulded to specification. Through the implementation of intelligent technologies, we aim to optimise the circularity of plastics, contributing to a sustainable and zero-waste future.
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know