Jessica Bradley speaks to industry leaders about how data is transforming waste operations.
AI technologies are transforming the way waste is processed. From AI-assisted route optimisation and sensor-led collections to digital waste tracking and automated facilities, data-driven technologies are reshaping waste management operations.
Many local authorities and waste management companies are taking the leap and investing in data-driven technologies, with the aim of achieving improved efficiency and stronger compliance as a result.
However, some limitations and costs remain, as well as the lingering question of what the future holds for humans working in waste management.
How AI is changing the waste industry’s daily operations
There are many ways real-time data is changing daily decision-making on the ground. For local authorities, this looks like improved insight into waste generation patterns and behaviour.
“The biggest task we use data for as a local authority is optimising waste collection regimes,” says Dave Atkinson, director of environmental and regulatory services at City of York council, a unitary authority with waste collection and disposal services.
Optimising its waste collection regime can bring numerous benefits for councils, including reduced operational costs, carbon footprint and traffic congestion, and increased recycling rates and improved service levels. But doing this without data-driven technologies can be hugely challenging.
“There’s so much data involved with waste because you have, say, 90,000 properties and hundreds of thousands of segments of road, and you’re trying to work out the best way of getting a [refuse collection] vehicle around them all,” Atkinson says.
AI is also helping businesses with Digital Waste Tracking (DWT), which requires huge operational shifts. Progress has been made since the government first announced mandatory DWT ambitions in 2018 as part of its Resources and Waste Strategy, which aimed to provide a ‘comprehensive way’ to see what’s happening to the UK’s waste.
The legislation focused on commercial and regulated waste activities, with operators expected to record information for each movement into permitted sites.
In February this year, Defra confirmed the mandatory rollout for phase two of DWT will take place in October 2027. This means reporting through DWT will become compulsory for all required operators across the waste supply chain.
The Rail Safety and Standards Board (RSSB) has been working with engineering and environmental consultancy Ricardo to help the industry transition. It has developed metrics to help Britain’s rail industry measure performance across the circular economy, waste management, and resource management.
Both organisations are now working with rail firms to pilot the implementation of data collection plans and metrics for the industry’s waste management. The RSSB says the information gathered using these metrics would help to build a data-led understanding of rail’s sustainability credentials, by enabling consistent monitoring and reporting of circular performance across its assets, infrastructure and operations.
How data helps decision-making and safety
Better data can enable better managerial and operational decisions within the waste management sector, according to researchers.
In a paper published in 2024, researchers outlined in the journal Cleaner Waste Systems how they used machine learning models – a subset of AI – to analyse and forecast waste generation trends. They also used it to assess the viability of numerous waste management methods and develop optimisation models for resource allocation and operational efficiency.
Better data can enable better managerial and operational decisions within the waste management sector, according to researchers.
They achieved 85% accuracy on predictive analytics models for forecasting waste generation trends (they attribute this to the integration of more diverse data sets) and a 15% increase in operational efficiency. They said their findings prove that machine learning models can lead to more sustainable and cost-effective practices.
Tom Harrison, sales manager at Recycleye, has seen firsthand how data-driven technology sharpens decision-making. Recycleye’s machines – a robot arm and an optical sorter – can be implemented to give operators a better understanding of preventative maintenance, he says.
“They see minute-by-minute data to get an instant understanding of what’s going on, including when something is going to break. Having the data is really useful for them to carry out active, rather than planned, maintenance,” he says.
Data also helps with financial planning, Harrison says.
“They can see if plastic content is higher one month, so they can see the long-term view as to whether it’s worth investing in different equipment.”
One downside of AI, Harrison explains, is that it can’t weigh material. When it comes to sampling waste, however, it’s possible to know how much items roughly weigh and extrapolate this to understand a waste stream and make informed decisions about it, he says.
However, there are many instances where human judgment is still crucial, and can complement data-driven technologies.
UK startup LitterCam uses CCTV footage to detect low-lying litter to help reprofile road sweeper and larger refuse collection vehicles’ routes. Its technology can also detect littering from vehicles.
“Data-driven approach enables insight-driven decision-making and ability to make decisions more quickly instead of relying on gut or doing things in the way they’ve always been done,” says Andrew Kemp, the company’s founder and chief executive. “The ability to make data-driven decisions is absolutely key.”
However, Kemp adds, local authority offices still have to validate suspected offence footage and look at appeals.
“They need to see the littering offence themselves,” he says.
Atkinson continues that the enduring need for human judgement in waste management manifests in several ways, but that the role humans play will shift.
“There is always going to be a need for mechanical type maintenance,” he says. “Particularly with complex machinery and robotics.”
“It’s not a stretch of the imagination to think that, in the next 15 to 20 years, we’ll have autonomous vehicles doing waste collection. But there’s a care element where members of the public would want human interaction between the council and members of the public. There’s also a safety element. We would want an overview, whether remotely or on-site.”
Harrison says there’s a safety element to having human input for Recycleye’s customer base, too.
Recycleye machines would, he says, never go at the pre-treatment front end of a plant, where waste is pre-sorted to remove any items that are dangerous or too big for the automated part of the process.
“I don’t think we will ever replace that side of it with machinery because you need that human understanding of what’s dangerous,” he says. “But once that’s all been treated, there’s nothing to say that, after that, you couldn’t have a fully automated production line.”
Challenges and barriers of data-driven technology
One of the limitations with AI is that it will never be 100% accurate, says Harrison.
“You train AI to see all the different products that come through the sorting line, but if it’s never seen something before, it has to make a prediction, which may be incorrect,” he says.
Danielle Stephens is the founder of Recycle Lab, a recycling start-up that collects and recycles plastic waste from science labs. She says cost is the biggest barrier for start-ups wanting to implement data-driven technologies.
Stephens says cost is the biggest barrier for start-ups wanting to implement data-driven technologies.
The company picks up waste from larger customers, while smaller customers’ waste is picked up by a courier. Stephens says that as the company grows, she hopes to look into AI-assisted route optimising – particularly because customers are using Recycle Lab as a way to be more sustainable.
“I started the business because of a lack of sustainable options within the industry, so we try to be as sustainable as possible,” Stephens continues.
“Route optimisation is a sales aid to show customers that we’re using this technology to help reduce our carbon impact, which in turn helps them to reduce their carbon emissions.”
Currently, Recycle Lab manually tracks collections and sends customers data about their waste, but says this will become increasingly difficult as the company grows, and it will need to embed DWT.
Cost is the biggest reason Stephens hasn’t invested in these technologies yet.
“In the next five years, we’ll also be looking at how we can improve efficiency through automation and machinery, but first we need to have more capital to warrant the investment,” she says.
Another challenge for waste management within the science industry, Stephens says, is legislation.
Because plastic waste and recycling are quite new in the science industry, Stephens explains that legislation is yet to catch up – and this means there aren’t yet enough standards or specific legislation around recycling.
The future of data-driven technologies
Atkinson would like York City Council to start automating the brokerage of materials to markets.
“The materials we collect have value, and it’s in the local authority’s interest to limit the contamination of recyclable materials,” he says.
He’d also like to introduce a large language model to start collecting data from social media showing customer feedback regarding any missed bin collections, for example, that could feed info through the periodic optimisation processes.
“This could transform council services to be more suited to what residents want,” he says.
On a wider scale, Harrison says data-driven technologies will continue to become smarter in the waste management sector – but that this is dependent on people.
“The model is going to get better, as is the nature of AI, which is constantly learning,” he says.
“It doesn’t learn by itself, though – we teach it what individual items are. Behind all these models are people.”
