TOMRA launches AI platform with natural language interface, three deep learning sorting applications

USA – TOMRA Recycling has unveiled a new AI platform from PolyPerception with natural language interface, plus three GAINnext deep learning applications for food-grade PET trays, copper-steel composites, and used beverage can aluminium recovery.

The PolyPerception platform evolves its AI-powered Waste Analyser solution, designed to improve sorting performance through end-to-end material tracking. 

Its natural language interface allows operators to ask questions such as “How did changing the settings on the recovery line affect our purity?” in plain language, with the platform providing immediate natural language answers accompanied by data breakdowns. 

The platform also has writing capabilities, enabling it to actively create custom quality reports and set operational alerts based on deep domain knowledge of the recycling process.

Smarter Sortation for Packaging Streams

The first GAINnext application addresses rising demand for food-grade PET trays, distinguishing between takeaway or supermarket trays and consumer or medical packaging based on shape and use, reportedly achieving purity levels over 95%. 

For PET recyclers, this capability is critical because tray materials differ from bottle-grade PET in molecular weight and crystallisation behaviour. 

The second application targets “copper meatballs,” automatically identifying complex copper-steel composites such as motor armatures even in oxidised or dirty streams, helping recyclers upgrade rebar-grade scrap to premium furnace feedstock. 

The third application is a high-throughput solution for used beverage can aluminium recovery from packaging streams, launched in North America and now adapted for the European market, offering up to 33 times more throughput than manual sorting while delivering 98% purity or higher.

Real-Time Problem Detection

The platform features two search methods to help plants respond to changing material streams. 

With similarity search, operators can right-click a problematic object to instantly identify every other visually similar item in the stream, which can be used for spotting fire hazards like batteries without needing to train a new AI model. 

Through text and brand search, users can search for specific brands or object types to see what is passing through the facility in real time.

When AI Talks to the Sortation Line

A sorting line that cannot explain why purity dropped is a guessing game. 

TOMRA’s new AI platform eliminates the guesswork, replacing manual log reviews with natural language queries.

For recyclers, that means faster troubleshooting, higher output quality, and less material downgraded or rejected.

Subscribe to our email newsletters that provide busy executives like you with the latest news insights and trends from Africa and the World. SUBSCRIBE HERE