Artificial intelligence helps analyse sorting quality


Fost Plus is developing an artificial intelligence pilot project in the PMD sorting centres. It is being put in place in partnership with the Indaver sorting centre in Willebroek. With the use of this technology, we want to assess the sorting quality in the sorting centres. 

Technology at the service of people

The sorting process is automated, but human resources are used to analyse the sorting quality, on the basis of samples. The quality of the recycled materials depends on that of the sorted bales which we deliver together with our operating partners. By noting any inconsistencies and optimising the flows of sorted materials, we can recycle more and provide the recycling centres with a constant supply of quality materials. Fost Plus and Indaver set up the artificial intelligence project, which measures the quality of the sorted bales, precisely to achieve this aim. The purpose of the technology is to supplement the human processing capacities for quality analyses and improve monitoring of the quality of the sorted material.

It is essential to involve the sorting centres in this pilot project. That is why Indaver was asked to collaborate on this project. As a matter of fact, as Eric Goddaert, Recycling Director at Indaver, explains: “Indaver has been looking into the possibility of integrating artificial intelligence into its production processes for some time, because the quality of the sorted materials is of the utmost importance. When Fost Plus suggested that we take part in the project, we didn’t have to think about it for long before agreeing.”

Detection of sorting errors  

A camera has been installed and is used on the sorting lines. It is connected to a software program that should be capable of scanning and detecting sorting errors at the final stage of the sorting process. Ultimately, the information obtained could even be used to improve the calibration of the sorting systems. 

The device installed takes one photo per minute and sends it to a digital portal. This collects all the images taken. Artificial intelligence automatically identifies the impurities in the sorted packaging waste which may be overlooked by the sorting system. This way, we can test the artificial intelligence model and its capacity to recognise the packaging and objects in each photo. At the moment, this trial is limited to identifying opaque PET (which is not allowed) in the clear PET flow. 

The use of artificial intelligence in the sorting sector is vital. “At Fost Plus, innovation and digitisation are very important in the context of the circular economy”, says Philippe Alen, Quality Manager at Fost Plus. “Technology helps us to closely monitor the processes used and make adjustments if necessary. Good quality control guarantees that we continue to meet high standards. The sector in which we operate is constantly changing. So we have to keep innovating in order to respond properly to these changes.”

Continuous improvement

At the heart of this project, we still face a number of challenges for the use of artificial intelligence in the sorting sector. “We have noted that we still have certain problems to resolve before we can actually allow the project to embark upon a final phase. For example, we have worked on the connectivity to the platform in the cloud, and on the lighting available to photograph packaging. We will now continue to work with these observations.”, says Shanna Cloet, Project Leader and Business Analyst at Fost Plus.