Can AI solve the global waste problem?

With pressure growing to make the economy more circular, waste streams offer huge potential to turn a problem into a solution by considering them a valuable raw material for all kinds of products and processes.

But the sheer volume of waste, coupled with collection infrastructures and sorting facilities that are globally very patchy in existence and in quality, pose a serious challenge. Artificial intelligence (AI), particularly when combined with technologies such as automation and robotics, has huge potential as part of the solution to help make waste management more efficient and more circular.

The United Nations Environment Programme’s (UNEP) Global Waste Management Outlook 2024 report says that increasing circularity in waste management could save more than USD 100 billion a year by 2050.

Research shows that AI can drive waste management efficiencies by:

• Reducing the distance traveled by waste trucks by up to 36.8%

• Cutting costs by up to 13.35%

• Cutting the amount of time spent on collection by more than 25%

The figures are impressive, so let’s take a closer look at the opportunities, the potential and the challenges with AI being deployed to harness waste.

Plastics

Join the change towards a more sustainable plastics industry

Learn more today

From waste to high-quality raw materials

With waste management it isn’t just about quantity and volume – it is also about being able to recycle and process very heterogeneous waste streams. Currently, many waste streams are extremely varied, mixing for example paper and organic waste and containing multi-product packaging – and this type of mixed waste is difficult to process. Treatment facilities face dealing with these ever-changing waste streams containing everything from fishing nets to lithium-ion batteries.

Within waste management facilities, AI can make identifying and sorting waste more accurate, diverting it from incineration – which together with landfill should be considered a last resort. Operators are now, for example, installing cameras on waste-sorting conveyor belts and using AI to offer a precise breakdown of the materials in each stream and bale.

”Identifying different waste streams is an obvious area where AI will help,” confirms Sami Ahma-aho, AI Architecture Lead at Neste, the leading producer of renewable diesel and sustainable aviation fuel (SAF) and one of the leading suppliers of renewable and recycled feedstocks for plastics manufacturing.

Furthermore, increased data and digitalization offer the possibility of strengthening the recycling value chain. AI is one of the promising tools to help make waste management more circular and increase the amount of materials that can be recovered, with potential applications ranging from smart bins to waste-sorting robots. At the same time, AI can also help when it comes to keeping track of waste.

”Tracking and tracing different types of raw materials is important for companies like us that want to verify the origin and sustainability of the raw materials we use, which are mostly waste and residues. AI will make that easier,” says Ahma-aho. 

Changing behaviors with AI

Additionally, AI can help identify what types of items and materials end up as waste. “One of the first applications of AI in waste management will be in shedding more light on consumer behavior in relation to plastic waste,” says Ahma-aho.

A better understanding of waste streams, and how they are affected by consumer and businesses behavior, will enable the creation of strategies to help change that behavior. “That knowledge will show how we need to set up waste management infrastructure differently,” he says.

A U.S.-based company Clean Robotics is applying this principle to individual AI-enabled smart bins in high-traffic commercial spaces such as airports, stadiums and convention centers. “It is impossible to keep up with all the rules on recycling, and they’re different everywhere you go,” says the company’s co-founder and chief technology officer, Tanner Cook. “These spaces traditionally have very high volumes of waste and very low rates of recycling. Our products can be programmed to reflect local regulations and be updated when the rules change.”

Clean Robotics’ “trashbots” use AI and machine learning to identify items that can be recycled but also to educate consumers not just on what items can be recycled but also how they can help with that process, for example by emptying bottles and cups before throwing them away. The impact can be significant, Cook says: “Each individual trashbot can increase the quantity of recycled materials by about 1 ton a year.”

Are we ready for AI to help solve the waste problem?

One of the main challenges in integrating AI into waste management facilities are the limitations of current sorting infrastructures. Introducing AI for sorting can reduce the throughput of processed waste by as much as 50%, depending on the level of detail. The AI itself is really fast, but converting its instructions into action requires the material to be fed through the sorting station several times to get the required composition, or have several sorting stations, all of which increases costs.

Nonetheless, once AI has been successfully integrated with existing sorting technologies so it can handle increased capacity, it will give waste processors a new level of control, which will make it easier to provide increasingly varied and higher volume of waste-based raw materials.

Cook adds: “AI and machine learning will be important, but they have to work with other complementary technologies to be useful – robotic systems and sensors make it possible to deploy AI successfully and to maximum effect.”

Credits:

Mike Scott, An award-winning business and environmental journalist whose work has appeared in publications including the Financial Times, the Guardian and Forbes.