Urbaser plant unlocks new sorting capabilities with AI robot

The Algimia MSW treatment plant has reported positive results achieved by the AI robot installed over the residue line in June 2024.

The plant belongs to the C3/V1 Consortium of the Community of Valencia, and is managed by Reciclados Palancia Belcaire, a company of the Urbaser group.

The plant’s operations team observed that there were enough recyclables being left on the residue line to justify investment in equipment to collect and separate the valuable materials. The aim was to increase efficiency of the treatment process, lowering costs of sorting whilst extracting increased value, aswell as reducing the amount of materials going to landfill unnecessarily and thus reap tax savings.

Hence, Urbaser invested in an AI-powered sorting robot that was quickly and easily retrofitted the existing reject belt at the Algimia plant in June 2024. The robot, known as Recycleye QualiBot®, was designed and installed by British technology company Recycleye, with 45 similar systems already operating in sorting facilities across Europe.

The intelligent robotic picker positively picks missed recyclables from the residue line, sorting into 4 separate categories: PET bottles, HDPE containers, aluminium cans and paper and cardboard. It achieves this by using a computer vision system to scan each item on the belt, identifying each by object and material (e.g. aluminium can) using visual features taught by a machine learning algorithm trained on a dataset of 1 billion images. The AI system then directs the robotic arm to perform the physical task of picking the item and shooting it into the correct bin.

This novel detection method has unlocked new capabilities for the Urbaser team, as well as increasing consistency of existing picking tasks continuously throughout the shift. For example, the robot picks uncrushed or liquid-filled PET bottles which traditional optical sorters either struggle to detect, or struggle to eject because they are rolling. Additionally, the robot has been more successful at separating heavy pieces of paper, such as magazines and books, increasing the recovery of paper and cardboard from the line. It also detects black HDPE containers accurately on the black conveyor belt, which have previously been a blind spot for NIR optical sorters.

The AI’s ability to sort waste by type of object, not just material, has allowed the Urbaser team to extract exclusively aluminium beverage cans, and leave lower-value aluminium objects, such as foils or trays which often contain unwanted organics, on the residue belt. This has bolstered the purity and value of the aluminium can fraction extracted.

Thus, the Recycleye QualiBot® has increased consistency recovery of recyclables from the residue line, reducing landfill costs and increasing plant efficiency, whilst also producing a purer, and thus higher value, plant output.

The robot is already recovering an average of 50kg of target material per hour, enabled through a pick rate of 1500 picks per hour, consistently higher than the speed originally agreed with the client.