Before addressing the challenges, it is important to understand the technology and basic concepts of artificial intelligence and computer vision. Although it may seem like a recent development, AI has existed for many decades—the term was first coined in 1956—and is present in many aspects of our daily lives.
What do we mean by Artificial Intelligence (AI)?
AI is the science that seeks to replicate human intelligence and its processes through computers. In other words, it involves giving machines the ability to reason and make decisions using algorithms, enabling them to perform tasks traditionally carried out by humans.
Within AI, we find Machine Learning (ML) mechanisms. ML is a branch of AI based on systems learning from data, identifying patterns, and making decisions with minimal or no human intervention—limited mainly to the training phase. Common applications include predictive models, virtual assistants, and process optimization tools.
Examples of ML methods include decision trees, often used in predictive modelling, and linear discriminant analysis, applied in pattern recognition.
Nested within ML, like Russian matryoshka dolls, is Deep Learning (DL). DL is characterised by algorithms capable of independently learning to analyse and interpret data inputs. Put simply, these algorithms allow machines to learn in a way similar to a small child—improving their performance through experience without human intervention.
Why is AI experiencing exponential growth?
Several factors have contributed to this rapid development. Firstly, there is a growing need to process and analyse vast amounts of data quickly. Secondly, these algorithms are highly adaptable to almost any process. Thirdly, hardware has evolved to become more powerful and affordable, without requiring large physical space. Combined with an increasingly digital society, these factors have created the perfect environment for AI to grow exponentially and become present in nearly every sector.
How is this technology being used in the recycling sector?
The most straightforward example in recycling is in waste sorting, particularly for packaging. The demand for higher-quality recycled materials is increasing, making more precise sorting essential to ensure a good final product. AI, combined with computer vision and NIR cameras, enables outcomes such as streams composed exclusively of PET water bottles, the separation of construction waste by type, or sorting garments by material and fabric type.
These automated sorting processes are also supported by big data and the Internet of Things (IoT), allowing real-time monitoring and control. This enables immediate action in the event of errors or equipment failures.
Another application is in waste management systems, where AI and IoT can significantly optimise collection processes. For example, sensors can detect when urban waste containers are full, allowing collection routes to be adjusted so trucks only visit full containers—saving time and transport costs. However, one barrier to widespread implementation is the high cost of these technologies.
What does the future hold for AI in plastic recycling?
Considering current applications and the opportunities offered by DL algorithm development, we can expect optimal recycling processes, with output streams free of impurities and higher-value final products. AI could identify specific objects that may cause issues in later recycling stages and connect the entire value chain—from citizens to recyclers—to improve communication and efficiency.
A more distant possibility is the separation of waste fractions with specific value, such as food-contact packaging, where recycled material use is limited due to the difficulty of obtaining exclusive waste streams of this type.
Conclusions
AI is a versatile and increasingly powerful tool, driven by hardware evolution and digitalisation. Its rapid growth allows it to be applied across diverse industries. In the recycling sector, it offers great potential to improve and optimise current processes, achieving cleaner and higher-value output streams by integrating these technologies throughout the value chain.
