MACHINERY
Artificial intelligence is revolutionising the optical sorting of materials, opening up new avenues for recyclers.
Stefano Bonacina, sales director of 3U Vision, explains the company‘ s strategies for innovation, specialisation and new markets in order to compete in an increasingly challenging market.
AI OPENS UP NEW HORIZONS IN SORTING
Italian technology plast / October 2026
038
The optical sorting sector is experiencing a pervasive change. Artificial intelligence is changing the way materials are identified and separated, while the evolution of recycling is creating fresh opportunities far beyond plastic. This is the vision of Stefano Bonacina, Sales Director of 3U Vision, who explains how the company is going through a period of significant technological and competitive change. The starting point remains the recycling of plastics, but the scenario is far from simple. According to Bonacina, the sector continues to suffer from the lack of competitiveness of recycled materials compared to virgin resins, a situation that has been slowing investment for several years. In addition to this, there is a still fragmented regulatory framework where tools such as deposit-return systems could help increase the quantity and quality of collected waste, fostering a true circular economy.
OPTICAL SORTING IS CHANGING If until a few years ago, a sorting machine was primarily based on cameras and optical sensors, today the scenario is completely different. „ Artificial intelligence is turning the tables,“ explains Bonacina. „ We are no longer limiting ourselves to separating materials by colour or optical spectrum. Today, we can train models capable of recognising seemingly identical objects intended for different applications, such as a bottle for food and a bottle for non-food.“ The real challenge is not so much the algorithm itself, which is destined to become increasingly accessible, but rather creating models using huge amounts of correctly classified data. 3U Vision is focusing a significant portion of its technological development efforts precisely on the management of big data and the training of AI systems. The goal is to offer increasingly accurate systems for classify-
www. plastmagazine. it