Smart Waste Management: Leveraging AI and Robotics for Better Recycling
The waste management industry is adopting waste sorting technologies that rely on artificial Intelligence (AI) to tackle the growing problem of municipal waste. The UN’s Global Waste Management Outlook 2024 report, from their Environment Program (UNEP), says that municipal solid waste production will grow from 2.3 billion tonnes in 2023 to 3.8 billion tonnes by 2050. This is due to more housing being developed in cities leading to more domestic waste, largely made up of the packaging used for food and beverage products and for online shopping.
AI promises to improve waste sorting
AI systems help to improve the automation of waste sorting by making it more accurate and productive. Waste management companies need waste sorting solutions that can handle high volumes of wastes. Waste streams that enter recycling, particularly in the EU, are often a mix of materials including paper, card, glass, metals, plastics and an increasing amount of flexible packaging. The aim of recycling is to separate the different materials so that they are at least 95% pure, a challenging goal considering the diversity of input materials. AI can be used to separate off impurities, or select specific packaging types to yield purer recycled materials that command higher prices and are more easily used as raw materials.
AI reduces the need for manual picking by human operators
Improvements to automated sorting also reduces the need for manual separation of waste. Many materials recovery facilities (MRFs) have staff that work on picking lines to sort waste. These operators hand pick through the waste to remove unwanted or non-recyclable materials, e.g., pots and pans, contaminated packaging and unsuitable materials, such as wood. The use of AI systems may reduce the need for manual picking of waste by staff. This should reduce the need for these unpopular and risky jobs where the repetitive physical tasks and close proximity of humans to machinery makes the working environment hazardous.
AI-powered object recognition
AI systems use machine learning and computer vision to identify and categorize waste materials. Images of waste materials are taken by cameras and sensors as the waste moves along conveyor belts and these images are then analysed by computer vision. Wastes are recognised by identifying features within the images such as the shape, size, colour and texture of the objects. Scottish startup, Danu Robotics, claims that the accuracy of their computer vision system is because it identifies objects by their shape and edges rather than by drawing a rectangular bounding box around each object. Machine learning models compare the selected features against a training database of known waste materials to identify and classify each item. Some of these databases are huge, e.g., UK company Recycleye claims their WasteNet database has over three million training images, is the world’s largest dataset for waste, and can be used to discriminate objects based on weight as well as allowing brand-level detection. Their AI system can scan 100% of the objects on a conveyer belt, allowing waste processors to continually monitor the composition of their waste streams.
Integrating AI with waste sorting
Several factors will need to be considered when deciding to purchase an AI-enabled waste sorting system. The options include modifying existing waste sorting equipment, expanding an MRF site, or building a new facility. Waste sorting equipment wears out, so retrofitting in an AI system may only make economic sense if the current equipment is less than ten years old. The availability of physical space and electric power may also need to be assessed. There are two main methods for extracting the target waste materials from a conveyer belt, robotic arms and positive pressure air ejectors. Robotics are considered a good solution for retrofitting especially if space is limited, but air ejectors can pick ten times faster than robotic arms and are often more effective.
Waste streams that can be purified using AI
AI systems can be used to sort and enhance the purity of a variety of waste streams. Some can handle a broad range of materials e.g., German company, Tomra, supplies high-throughput sorting systems that can separate and purify paper, metals and plastics (PET, PP, PE) from post-consumer packaging streams. Waste Robotics, a Canadian company, uses a hyperspectral camera to improve waste sorting and discriminate between different types of plastic including PE and PP. MSS Vivid AI, developed by US company MSS in collaboration with Recycleye, can detect subtle differences in object color, shape and texture and separate used beverage cans (UBCs) from non-UBC aluminum. New applications are being developed, SWEEEP Kuusakoski, part of Finish company, Kuuskoski Oy, collaborated with Recycleye to use AI-powered sorting to remove hazardous lithium-ion batteries from a commercial waste electrical and electronic equipment (WEEE) stream.
The market for AI-waste sorting systems
The market for AI-waste sorting systems is definitely growing, but it is hard to get a good understanding of market size and growth rate. The published data are highly variable, with several generalist, market research companies providing figures for the market growth rate for robotic and AI waste sorting systems as anywhere between 6% and 20%. Figures for the current market size fall between $1.5Bn and $13Bn with a median value across five reports of $2.2Bn (SAL analysis). By contrast, in their 2024 pitch deck for seed funding, Danu Robotics stated that the total addressable market for automated waste sorting for household and packaging waste is much larger at $88Bn. Danu have priced their H.E.R.O. Model One, two robot arm picker system at £60,000 (approx. $78,000) with a monthly subscription fee of £1,500 ($1,945). The number of recycling facilities in the USA is on the order of 1,050 with the EU having far more, published data say 200,000+ facilities. If only 10% of these upgrade to add in AI-waste sorting, the market easily reaches $1.5Bn (using the published price for Danu’s system).
Benefits to the environment of improving recycling
The UNEP says we need to turn rubbish into a resource. Waste contributes to the climate crisis by being a source of the greenhouse gases, CO2 (particularly if rubbish is burnt) and methane (CH4). Litter pollutes water courses and leaches hazardous chemicals and if uncontrolled, is a driver of biodiversity loss. From an economic perspective it makes sense to recycle high value waste items like the scrap metals, steel and aluminium, and also glass. Secondary aluminium production also has the benefit of a five- to 25-times lower carbon footprint than primary production methods. Recycling also reduces the cost and environmental burden of creating and maintaining landfill sites. The UN say “a circular economy and taking a zero-waste approach is the only route to a safe, affordable and sustainable future”.
Conclusion
The addition of AI to waste sorting processes can have several benefits. Robotic sorting systems can reduce the need to employ staff to do waste picking and help to make the industry safer. AI systems can improve the purity of waste streams which is particularly useful in secondary metal recycling, where the price of recyclates varies depending on the type, quality, and grade of materials, so accurate picking and sorting are needed. Strategic Allies Ltd is keen to discuss with you how AI-powered sorting could bring benefits to your business as we know it is also used in the food and beverage and pharmaceutical industries. Is there another sector where it could be applied? If you are interested in exploring these opportunities, please reach out to John Allies at john@strategicallies.co.uk for an initial conversation.