Calls for a worldwide transition to a circular economy are growing louder as the effects of climate change become seemingly more apparent with each passing year. The spotlight on action rather than words is spurring innovation in sorting waste materials ready for recycling in a cost-effective and process-efficient manner. Among the promising solutions to this challenge is the use of waste-sorting robots with computer vision and artificial intelligence (AI) capability.
Plastic, as most are all-too aware, is a problem material. In 2022, the OECD reported that only 9% of plastic has ever been recycled. Due to the critical nature and urgency of addressing this environmentally damaging dynamic, a tremendous business opportunity is afoot.
“The amount of plastic production goes up every year, as does the cost of recycled plastic,” states Peter Hedley, CTO at Recycleye, a technology company bringing machine learning, computer vision and robotics to the global waste management industry. “That’s really building this marketplace; waste sorting plants and MRFs [material recycling facilities] need high-quality sorting and automation to deal with the load. We believe that waste does not exist, only materials in the wrong place.”
Recycleye co-founder and CEO Victor Dewulf adds: “This is a market which we estimate to have a serviceable addressable market value of $114 billion globally today, but with the potential to increase 14 times to $1.6 trillion when the cost of sorting is reduced.”
ROLE MODELS
Using proprietary AI models, Recycleye’s robotic technology ‘sees’ waste and can pick an unlimited number of material classes – not only plastics, but also aluminium, paper and cardboard. It can even differentiate between packaging and non-packaging materials, and colours and shapes.
Nick Kastanos, machine learning engineer at Recycleye, says: “Aside from simply identifying an aluminium beverage can, for example, the AI algorithm can also learn what a coffee cup is, or a plastic cup. This way, the same robot delivers multiple points of value to the waste sorting plant.”
Adding yet more value is a recent development that focuses on expanding Recycleye’s AI-powered waste picking solutions with a new set of proprietary data. As a result, the company’s robotics solution is not only able to ‘see’ waste, but also to ‘smell’ it using a special sensor located in the gripper. Recycleye expects initial applications to include detecting the difference between food and non-food packaging, which of course usually smells either of food or chemicals, bringing yet more precision to the sorting process.
PURE AND SIMPLE
Increased accuracy on sorting lines equates to higher purity outputs, which has a direct effect on the revenue that MRFs can generate. With Recycleye robotics it is possible to scan and identify objects at 60 frames per second, which means that the robot ‘sees’ each item 30 times on average as it passes along the conveyor belt, enhancing the potential of accurate identification before picking. Each robot can pick up to 3,300 items an hour, while capturing compositional data to enable strategic decisions by plant managers.
Among early adopters is re3, a partnership between waste management company FCC Environment and Bracknell Forest, Reading and Wokingham Borough Councils. At re3’s MRF in Reading, a Recycleye AI-based robotic waste-picking system sits atop an existing split-belt conveyor, providing the site with visibility on its waste streams and enabling re3 to improve the efficiency of its sorting processes.
Rory Brien, general manager of FCC for re3, says: “We wanted better material purity on our plastics line as it helps achieve sustainable offtake. The Recycleye robotics system ensures that we keep HDPE Natural on one side of the split belt, before making sure it is free of contamination by removing fibres for instance, while the remaining mixed plastics stay on the other side. We’re getting fantastic results. Based on information provided by the Recycleye dashboard, the Reading MRF sees HDPE Natural with less than 1% contamination. In addition, we’re achieving a 12% increase in the volume of target material that the robotic system is picking from the belt, saving on processing costs.”
Another company helping the waste sector to improve its sorting capabilities is US-headquartered AMP Robotics, for which the official reseller and integrator in the UK and Ireland is REP-TEC Advanced Technologies. AMP is introducing AI developments that increase recovery performance and reliability. The company’s latest AI ‘Advanced Targeting’ algorithms use machine learning to determine the optimal grip region for each waste item based on the object’s discrete material features and condition. This ability will increase yield by learning to avoid creases, holes and other difficult-to-grip object locations. AMP’s AI-driven software advancements will be available for its Cortex and Cortex-C units.
Cortex and Cortex-C are among AMP’s portfolio of robotic recycling solutions powered by its neural network, which the company says recognises more than 75 billion containers and packaging types in real-world conditions annually. AMP Vision is a computer vision system that helps operators understand material flow throughout key stages of sorting operations. When integrated with AMP Clarity - the company’s portal for recycling data, insights and robot optimisation - users can monitor real-time material characterisation and performance across a facility.
Clearly, AI is fast becoming an authority on identifying and picking waste from conveyors, much to the benefit of the recycling industry, society and the planet.
BOX: NEVER TIRE OF AI
Giving new life to old tyres, reducing CO2 emissions and making a sustainable contribution to the circular economy are all goals that are said to be pursued by Zeppelin Systems. Together with international partners and using what it describes as innovative technologies, the company is now providing industrial recycling solutions for used tyres. French company Regom, which specialises in sorting used tyres using AI, has been on board since February 2023.
Every three seconds, an automatic sorting machine determines the future of a used car tyre; technology from Regom manages around 800 tyres an hour. This rapid rate is possible thanks to the use of AI which, after taking measurements, evaluates the side wall images of each old tyre at pace and determines its further use. The result: a highly accurate recognition rate of tyre brand, model, profile, size and wear.
“This process step is of great importance for the further processing and high-quality recycling of used tyres,” says Guido Veit, VP sales for polyolefins, rubber and silos at Zeppelin Systems. “At the same time, staff involved in sorting are relieved because tyres no longer require manual sorting and evaluation. Furthermore, due to the exact measurements and evaluation, the proportion of recyclable tyres automatically increases.”
Tyres that do not make it back to retail are subject to a subsequent step: mechanical separation and shredding. Thanks to high sorting precision, the quality of the resulting recyclates increases in terms of their purity. The upshot is the creation of a good starting point for the production of new tyres using recyclates from used tyres, thus supporting the circular economy.