Project OMNI’s AI-driven food-grade rPP research leads to “groundbreaking” results

24 Nov 2023 --- Project OMNI, a research initiative directed by Recycleye, Valorplast and TotalEnergies to enhance the circularity of PP food-grade packaging, has “led to groundbreaking results,” according to the group.

Project OMNI is an AI and machine learning project aimed at boosting the identification and separation of food-grade PP from household post-consumer waste. It is one of seven projects selected in October 2020 by Citeo.

The collaboration partners explain that after 18 months of research, Project OMNI led to an alternative to digital and physical marking solutions that require system-wide packaging changes. In a demonstration unit, Recycleye built and trained an AI model based on various forms of waste collected from five locations across France that Valorplast supplied.

The AI and robotic sorting achieved a successful pick rate of 50% of the food-grade material, with over 95% purity. This sorting activity produced material for further decontamination on a semi-industrial pilot based on off-the-shelf mechanical recycling technologies. TotalEnergies then leveraged its polymer expertise to produce odorless, clean rPP suitable for high-end packaging applications.

A Citeo spokesperson tells Packaging Insights: “The great thing about the use of AI to identify packaging waste is that it is a way to identify and sort packaging waste without system-wide changes, and so the packaging industry does not need to make changes.”

“The AI system developed by Recycleye is as accurate as a human eye, so if it can see the item on a belt, it can detect it, and it can be sorted. Being able to create a clean food PP packaging waste stream will enable the packaging industry to go further with the use of rPP in a close loop for packaging applications.”

Future steps, future challenges
While no additional phases are planned for the project yet, the team says it would like to repeat the work completed at a larger scale.

“Now that the feasibility of the technology has been proved, the intention is to go to an industrial level,” the spokesperson says.

“It is not easy to sort waste using AI and requires a large database of images to be able to identify the objects for sorting, so we required a large amount of waste, which was shipped from France to the UK, for scanning purposes.”

Waste brings an additional challenge in that each object has very different characteristics when it reaches a waste plant, compared to when the consumer first bought it (usually dirty, crushed, shredded or overlapping when it comes to a waste plant for sorting), they explain.

“AI identifies a particular object using its features, just as the human brain does, and so we are able to sort food grade PP by identifying the features of a PP object that usually contains food. If and when packaging changes, we would simply need more images of the waste to be able to identify and sort it.”

“We are excited by the application of Recycleye technology at this scale and for the results that the partners have been able to achieve by bringing their expertise together. The OMNI project is a perfect example of collaboration around the waste value chain,” the Citeo spokesperson concludes.

By Louis Gore-Langton

To contact our editorial team please email us at editorial@cnsmedia.com