AI as a catalyst for material circularity
The good news is that AI is now making this possible. What’s more, AI will become a catalyst for material circularity. Such advanced technologies significantly improve the sorting and classification of recyclable materials and help to meet the increasing demand for more recycled content.
Before we look at how AI is changing resource recovery, it’s worth clearing up a common misconception. AI is not just the latest buzzword but has always been at the heart of our industry. Our research and development teams have been developing AI-powered sorting solutions for years. Even our first TOMRA machines 30 years ago were able to make decisions about which material to sort into which container. And the ability to make decisions like a human is the very definition of AI.
When we talk about AI today, we are referring to the latest developments in the field of deep learning. This is a subcategory of machine learning that has found its way into our industry thanks to the advancement of computing power over the last decade.
1. Deep learning provides more flexibility. With the ever-changing composition of waste, sorting systems need to be agile enough to adapt to new market requirements. Instead of replacing hardware components or even machines, modern deep learning technologies can be retrofitted with regular software updates as soon as they have been trained by our experts. This allows technology suppliers to respond more quickly to customer needs.
2. Sorting itself will improve considerably. Conventional sorting systems have already achieved a remarkable degree of accuracy and effectiveness, such as advancements in separating single- and multi-layer PET trays. By combining existing optical sorting systems, which are based on near infrared (NIR) and/or visual information sensors (VIS), with deep learning technologies, as is the case with our AUTOSORT™ with GAINnext™, we achieve the highest sorting granularity currently available. This allows us to sort by material type and color and now, thanks to deep learning, also by shape, size, dimensions or other details. It solves previously impossible tasks, such as sorting food-grade vs. non-food-grade PET, PP and HDPE. This is a milestone for our industry, especially as GAINnext™ enables us to achieve the purity rate of over 95 percent required for food safety in Europe.
3. Deep learning will further advance the automation of plants. The value of deep learning lies in object recognition using full-color cameras. In other words, GAINnext™ sees what the human eye can see. We can automate sorting tasks that previously had to be carried out manually, enabling us to process larger quantities of recyclable materials quickly and efficiently.
4. Last but not least, we can take a big step toward process optimization with the help of data. AI-powered sorting systems generate huge amounts of data on material composition, sorting efficiency and plant performance. By analyzing this data, operators can identify optimization opportunities and act accordingly. Additionally, the possibilities reach beyond sorting systems. Cameras based on deep learning can be placed at key points in the sorting circuit to keep an eye on the entire process and material flow. This allows plant operators to continuously monitor the quality of sorted streams, material loss and even ensure compliance with food recycling regulations.
What is deep learning?
Deep learning is a subcategory of AI. It imitates the way the human brain processes information. It is a special technique within machine learning that uses artificial neural networks that are trained by huge amounts of data, so they recognize and store certain patterns and later apply them to new data. TOMRA’s AI experts feed thousands to millions of images into the network as training material until it learns to distinguish certain visual characteristics of material types such as specific bottle caps or packaging shapes. This enables deep learning to solve some of the most complex sorting tasks when combined with existing sensors.