Esegui ricerca
11 September 2023

Artificial intelligence for smarter recycling: the plastics pollution challenge

Aumenta dimensioni testoDiminuisci dimensioni testo

In 2019 only 9% of world plastics was recycled, and 50% still buried in landfills. With pollution growing dramatically, AI-powered robots are being tested to improve the sorting process and cut its carbon footprint. Experts stress the potential of the new technologies, but issue a call to industry players: “For this innovation to be effective, we need your support”.

When we throw a plastic bottle in the right waste container, we feel virtuous because we have done our part for the environment. However, we cannot imagine the huge effort that will be needed to ensure that our bottle is effectively recycled. After collecting, waste needs to be sorted, separated, and recycled into useful products, which sometimes risks costing more money and energy than producing new plastic. That is why only a fraction of plastic products is actually recycled. According to OECD estimates, in 2019 only 9% of world plastics was recycled, while 50% was buried in landfills and 22% mismanaged or left to pollute as litter. The amount of plastic already accumulated in the oceans, lakes, and river beds is staggering. Moreover, by 2060, global plastics use is projected to nearly triple from 2019 levels, driven by economic and population growth. 

Something needs to change, and fast. There is no easy solution to minimise the negative environmental effects of this scenario, but technological innovation to increase the production of secondary plastics can play a key role. “Boosting innovation for a more circular plastics life-cycle is critical” says Elista Lanzi, Senior Economist at the OECD and lead of the modelling team that developed the Global-Plastics-Outlook report, “yet until 2017 few patents were focused on waste prevention and recycling. Today, our patents analysis shows that innovation in this field is accelerating, also thanks to policies mandating ambitious recycling targets, such as the European Single-use Plastics Directive (SUPD)”. A recent JRC assessment highlighted the potential of emerging physical and chemical recycling technologies to produce high quality secondary plastics, with a carbon footprint 40-80 % lower than primary plastics. But they are not yet fully mature; they present several challenges and require a full transformation of the waste treatment industry. Several actors in Europe, such as the waste management industry, research institutes and universities, are collaborating to develop and test these new approaches. One example is the research project PLASTICE , which is running 4 pilots of advanced chemical recycling technologies. 

José Manuel González from COGERSA (public company for solid waste management in the Spanish region of Asturias), partner of PLASTICE, explains the need for better sorting processes for chemical recycling: “Today, with mechanical sorting, we normally achieve more or less 90-95% pure plastics (HDPE, PET, LDPE, etc.), but this is not enough: we need to achieve 95-98% purity of sorted waste for chemical recycling tests, so we need to improve drastically the sorting process. Moreover, robotics could help to recover valuable plastics from the refuse flows”. He adds: “That’s why we decided to join PLASTICE to test this robotic-based plastic sorting mechanism, which is expected to be faster and more accurate, because it uses special hyper-spectral cameras with better performance than human vision”. This technology can also replace low quality jobs with higher quality ones, as human intervention in the conveyor belt (a hard task in a dirty, smelly and noisy environment) will be reduced. 

AI (artificial intelligence) is the key to this innovative approach. The Greek ICCS (Institute of Communications and Computer Systems), partner of PLASTICE, is developing innovative industrial sorting systems for the Asturias pilot, which employ deep-learning models using data from advanced visual systems to identify all kinds of plastics and robotic arms to sort them. “We aim to increase the speed and accuracy of the process, so that the output of the waste treatment corresponds exactly to the specifications required by chemical recycling” say Fotios Konstantinidis and Savvas Sifnaios from ICCS. The institute is also working to apply this approach to other types of waste, such as construction and demolition, wood and mining. “The next challenge is bioplastic and metal scraps” adds Konstantinidis with enthusiasm. Higher purity materials can fetch better prices, which boosts the profitability of waste sorting. However, robots are costly.  “COGERSA will launch a new sorting plant for mixed waste in September 2023, with up-to-date technologies, but without robotics” explains González “because the current commercial solutions are expensive and don’t fit our needs. We hope this pilot will help us develop the right solution for us”. 

Finding a balance between innovation and cost-effectiveness is always difficult. This is where digital transformation, especially combined with AI, comes into play. Pedro Compais from CIRCE, the organisation leading PLASTICE, explains: “We are developing smart and digital solutions with the goal to enhance the recycling processes tested in the 4 pilots, leveraging data and artificial intelligence to optimise them”.  However, deep learning needs data to work and collecting the right data is easier to do at the laboratory scale, but more difficult at the industrial scale. In addition, researchers must compromise between robustness (to scale) and accuracy: if the model is too detailed and personalised for a specific use case, it is not easily replicated. 

Despite these challenges, in the future the costs of physical and chemical recycling are expected to decrease, while those for mechanical recycling (with all its shortcomings) will remain stable and virgin plastics production will become more expensive. According to JRC, chemical recycling will generate positive net earnings already by 2025, making it more appealing for mainstream adoption. “Given the seriousness of the situation of plastics pollution, this outlook is encouraging but will not be sufficient” says Elisa Lanzi from OECD. “We need more and faster innovation, social awareness (and there are signs of changing behaviour, for example, some people are switching to reusable bottles) and global collaboration for new policies. Europe can be a leader in this policy area, even though in absolute terms it produces less plastics waste than China or the Americas” concludes Lanzi. 

Figure 1 Elaboration on OECD Global Plastics Outlook Data provides its content to all media free of charge. We would appreciate if you could acknowledge as the source of the content.