Artificial Intelligence : A short history of AI in recycling
A fully circular economy relies on high-quality recovered materials, which current methods cannot achieve. AI, particularly Deep Learning, can be groundbreaking here. In material sorting with Deep Learning, a very recent development, neural networks are trained with vast amounts of images to distinguish visual features.
The many advantages of deep learning
By combining technologies like Near-Infrared (NIR) and visual sensors (VIS) with Deep Learning, materials can now be sorted not only by type and color but also by shape, size, and other visual characteristics.
This enables previously impossible sorting tasks, automates manual processes, and opens new business opportunities. Additionally, sorting systems can be updated via software to adapt to new market demands, allowing operators to respond flexibly to changing waste compositions.
The industry's first Deep Learning system
In 2019, industry leader TOMRA introduced the first Deep Learning system: GAINnext™. While the innovation was initially available for only one sorting task, today there is an entire ecosystem of applications.
GAINnext™ enhances material streams by removing hard-to-classify materials like colored paper, envelopes, and receipts in paper sorting, or opaque white bottles, textiles, multilayer films, and full sleeves from PET streams.
The technology peaked in 2024 with the first industry-wide solution for sorting food and non-food PET, PP and HDPE packaging. Recently, TOMRA announced the introduction of GAINnext™ in metal sorting to further improve the quality of recycled wrought aluminum scrap.
AI: A catalyst for green transformation
With stricter regulations and growing consumer demands, our industry is at a turning point. Deep Learning has the potential to significantly advance the circular economy and create new markets for higher-value products, and it's happening at the right time.
Learn more about the first and leading Deep Learning solution, GAINnext™.