“AI-enabled machine vision is needed to detect materials that current machineries cannot. Greyparrot software gives us data insights to maximise our material recovery and purity.”
Head of Innovation at A2A
As one of Italy’s largest multi-utility companies, A2A leads the country’s waste management and recycling efforts. To meet that national-scale challenge, they’ve partnered with Greyparrot.
A pilot project began in 2020, and has since evolved into a long-term partnership. In the effort to maximise material recovery and purity, A2A have deployed Greyparrot’s AI waste recognition system. They use the system to automate waste composition analysis at scale, giving them a better understanding of material recovery facility (MRF) operations and fluctuations in purity.
Guglielmo Carra, Innovation Manager at A2A, shares the story and the results of the pilot project — and why they’ve continued to work with Greyparrot.
With over 12,000 employees, A2A is a leader in Italy’s environmental sector. From waste collection to material and energy recovery (and everything in between), they operate according to a circular economy model that transforms waste into resources.
A2A’s pilot story
“As a company that collects and treats waste, a key point for us is to be able to extract the maximum value out of the waste that we manage.”
Innovation Manager at A2A
Without accurate sorting, it’s impossible to create resource flows that are homogenous enough for effective recycling. Material streams like paper and plastic need to be as free of contamination as possible to extract maximum value.
An efficient separation process is critical for businesses that operate MRFs and plastic recovery facilities (PRFs), like A2A. It’s easier said than done.
Those businesses face a huge lack of data and insight into waste composition. In order to sort waste accurately, they need to increase their understanding of the hundreds of tonnes of material that run through their plants on a daily basis.
Most recovery facilities currently rely on manual sampling to do so. It’s a costly, difficult to scale approach that only represents a snapshot of material composition.
AI counteracts the shortcomings of manual sampling. Instead of providing a snapshot, computer vision technology monitors 100% of the waste on moving conveyor belts. It then provides real-time data on object material, function, brand, and even SKU.
In doing so, AI can accurately differentiate different types of plastics, as well as food vs. non-food grade packaging from the same polymer.
“Thanks to Greyparrot we are able to get composition analysis done in continuum and in real time. Therefore we are able to create metrics and data sets that we can use to inform our decision-making on how to optimise our [waste] selection and sorting processes.”
Innovation Manager, A2A
Why A2A chose to partner with Greyparrot:
- The hardware - The A2A team were able to install Greyparrot monitoring units in their sorting facilities without having to adjust existing infrastructure.
- The software - Armed with real-time insight into their waste streams, A2A operations managers are now able to make data-driven adjustments to their processes, and to accurately report on product purity. The waste analytics dashboard allows individual facilities to customise metrics depending on their specific needs.
- The cost - When compared with an increased manual sorting effort, Greyparrot provides a scalable, sustainable means of gathering more data without undermining the value it offers.
During an initial proof-of-concept, the Greyparrot system reached an average weighted accuracy of over 95%. In simple terms, A2A received a clear and reliable understanding of the exact materials and products that passed through their conveyor belts — on a minute by minute basis.
The results also showed that data gathered through automated and continuous monitoring is a much better representation of waste composition compared to manually-gathered data. Continuous data showed that streams were far from homogenous, with purity fluctuating significantly every few minutes. A ‘spot-sampling’ approach risks ignoring that fluctuation entirely.
After a successful pilot, A2A agreed to a long-term partnership with Greyparrot. They are planning to expand the use of the AI monitoring unit to multiple product lines, in facilities across Italy.
“We are delighted to announce that our pilot with A2A has evolved into a long-term commercial partnership. We are looking forward to continuing our work together to keep increasing efficiency and transparency in waste management.”
CEO and Co-founder, Greyparrot
Greyparrot’s AI waste recognition system has been deployed at sorting facilities in over ten countries. By analysing billions of waste objects every year, it’s automating waste composition analysis and helping businesses like A2A to monitor, audit and sort waste flows at scale.
Greyparrot empowers waste managers, producers and regulators with actionable insights to:
- Increase recovery rates
- Monitor and adjust recovery processes
- Reduce the cost of manual sampling
- Improve product purity
- Guide packaging design and regulation
- Track the fate of waste objects from collection to landfill.
Greyparrot also collaborates with equipment suppliers to integrate cognitive intelligence with existing sorting machinery and software. An open API means that the platform is hardware-agnostic, and can be used to empower third-party technology like robotic arms, optical sorters, balers and more.
The company is committed to unlocking the financial value of waste. We believe that it’s a crucial part of our transition to a circular economy, and the larger effort to keep our planet clean for generations to come.
“The team of Greyparrot has always been very supportive and I’ve always been very positively impressed by the competence of the team and the vision that the team has for a better future for the waste industry.”Guglielmo Carra
Innovation Manager, A2A