How recyclables are sorted is becoming ever more important as waste streams become more diverse and contaminated. According to Victor Dewulf, ceo and co-founder of UK firm Recycleye, robotics are on hand to help. He shares his expertise on the topic.
For many companies the world over, 2020 was a rocky year. For Recycleye, it wasn’t all bad news as it achieved some important milestones:
- Closing an investment round of £1.2 million from leading venture capital firms and securing grants from the EU and UK government
- Building partnerships with Microsoft, Imperial College London, TU Delft University, the EU and others
- Scaling up the technology and R&D team from two to 14 people
- Deploying Recycleye Vision across the UK and France with several of the largest players in the waste industry
- Offering analytics software to give full visibility of a facility’s performance through a single dashboard interface
- Establishing a partnership with the world’s second largest robotics manufacturer, Fanuc
- Gaining publicity through the BBC World Service radio programme ‘Digital Planet’
- Co-founders were included in the 2021 edition of Forbes’ 30 under 30 Europe for Social Impact
Can you tell us the story behind Recycleye?
‘Before founding Recycleye, I resigned as a banker at Goldman Sachs in 2019 on a mission to remove the dull, dirty, and dangerous manual jobs plaguing the recycling industry. I first met our chief technical officer Peter Hedley at university and, after following different career paths for a few years, we came together to launch Recycleye. The company was spun out of my research as a PhD student at Imperial College London where version one of the Recycleye algorithm was developed. Then driving around the streets of Poole to collect household waste to run over a cheap treadmill purchased from eBay to refine the product.’
How many systems have been installed worldwide to date, and where?
‘Recycleye Vision is accurately identifying and sorting waste in recycling facilities across the UK, France and India with some of the world’s largest waste management players. We have installed more than 10 vision systems to identify household waste materials such as plastics, aluminium, paper and cardboard.
Our robotic picking system – the combination of Recycleye Vision with low-cost robotics – was developed in partnership with the Manufacturing Technology Centre (UK) and the world’s second largest robotics manufacturer (Fanuc). It has already been deployed at two UK material recovery facilities on multiple plastic and paper sorting lines. Deployments in France are scheduled for later this year.’
In what way did Fanuc’s expertise in robotics benefit the smart sorting solution?
‘Our partnership with Fanuc has been crucial in the optimisation, manufacture and roll-out of our robotics offering. Fanuc is a world leader in factory automation for CNC control systems, robots and production machinery. With 263 locations supporting 108 countries worldwide and more than 7 000 employees, Fanuc offers a dense network in sales, technical support, research & development, logistics and customer service. With their global presence and 60+ years’ experience in providing automated solutions for industry, our systems can now be installed and maintained across the world.
Fanuc’s team of expert automation engineers designed Recycleye Robotics to weigh 75% less than any existing robotic waste picker currently in the market. It performs the physical tasks of identifying, picking and placing material at a rate of 55 successful picks per minute. The novel solution automates manual operations and enables facilities to double their total throughput. Moreover, the plug-and-play installation eliminates the traditional cost of an expensive retrofit. Having installed more than 700 000 robots worldwide in different markets, Fanuc will utilise its expertise to help lead the global rollout of the Recycleye pickers, creating a new wave of intelligent and affordable automation in the waste management industry.’
What kind of material is Recycleye best suited for? Can it sort everything?
‘Our robotic picker can pick out all types of materials such as PET, HDPE, PP, aluminium and paper. At one deployment, Recycleye’s vision system is able to distinguish outgoing waste flows between food-grade HDPE plastic and non-food-grade HDPE plastic. Using computer vision to make this distinction enables facilities to provide higher value bales and subsequently leverage this data to charge their clients based on the individual composition of every tonne. Such detection is enabling MRFs to capture higher value from their outgoing waste flows.’
How has it evolved since it was first developed?
‘One of the main challenges we experienced was deploying Recycleye Vision to tackle different waste streams across Europe. As our clients were interested in sorting different materials, it was increasingly challenging to train our models effectively in a remote working setting. We overcame this by developing an automated pipeline that allows us to customise our computer vision algorithms for each customer and deploy them at the push of a button.
Another challenge we faced was dealing with huge volumes of data from the thousands of items being scanned by Recycleye systems in facilities every minute. Our system currently holds 270 million images – the world’s largest dataset of scanned waste. Recycleye’s dashboard platform has been key to managing all this data and providing plant managers with drilled-down metrics to enable them to fully optimise their plants and achieve the maximum throughput of waste sorting capacity.’
In your view, what is today’s biggest sorting problem?
‘The reliance on manual labour seems to be one of the most significant sorting problems as it prevents facilities from verifying the composition/contamination/purity of entire waste-streams and incurs large OPEX. Consequently, the cost of sorting waste often outweighs the perceived material value, rendering it cheaper to incinerate it or export it.
What would you say to sceptics who perhaps still doubt the potential of AI sorting solutions and the need for an upgrade?
‘To date, the limited availability of scalable recognition technologies has pushed the waste management industry towards a reliance on manual waste pickers to identify and extract high value materials. In recent years, however, the economic efficiency of manual sorting has been challenged for a number of reasons. Firstly, the average manual rate is 40-45 picks per minute. Although this may seem high, picks vary significantly depending on the length of shifts and operations such as shift changes and toilet breaks.
In comparison, Recycleye’s robotic picker does not require breaks or shift changes and works at a high speed with consistent output. Our robotic pickers exceed human performance with 55 successful picks per minute, enabling facilities to double their current throughput. With the UK having failed to hit its recycling target for 2020, automating the sorting process is the only way forward.
By embedding computer vision, recycling facilities are provided with 24/7 compositional waste item analysis on metrics such as item count, plant downtime and weight which give the traceability and accountability needed to accommodate European EPR policies.’
What is the potential of artificial intelligence-driven products? Have we as a society glimpsed only the tip of the iceberg – is the best still to come?
‘There is so much more to come. Computer vision is a low-cost and affordable solution to many of our global issues. In the context of recycling, computer vision is key to improving separation of waste to enable more materials to be recycled. In addition, using computer vision at different stages of the recycling process, such as intelligent bins or collection services, can not only increase recycling but also reduce the cost of recycling.’
You could argue that AI isn’t ‘perfect’ as it is only as smart as the people programming the system. What do you think can be improved to take it to the next level?
‘In the case of recycling, brand level recognition using computer vision introduces the transparency, traceability and accountability that the waste infrastructure is currently lacking. Moreover, our technology currently works in conjunction with optical sorters within the waste sorting infrastructure. This could be brought to the next level, though, by integrating it into other existing machinery such as air jet shooters and eddy currents. In turn, this would further maximise the waste sorting yield.’
What is Recycleye hoping to accomplish in 2021?
‘Recycleye’s unique capability of providing remote installations is helping to achieve our goal of capturing as much of the European market as possible: firstly with the sale of our robot and then, when licensed, with ongoing remote support and updates. The company will also be deploying its first two robotic pickers later this year – a significant milestone given development having being held remotely. Recycleye will also be undergoing its next investment round later this year, which will provide capital to expand not only our technological capabilities but also our commercial bandwidth.’
How do you envision smart sorting systems to impact the recycling industry in the future – say the next ten years?
‘Computer vision systems are necessary for moving MRFs towards a tech-enabled model where most operational decisions are made by machines, allowing facilities to adapt dynamically and rapidly to changing conditions, such as the pandemic. With the EU having defined its recycling targets of 70% by the year 2030, we envision every country in Europe hitting those targets with Recycleye’s technology!’
Would you like to share any interesting developments or article ideas with us? Don't hesitate to contact us.