Low-cost sensor developed by Imperial students to identify recyclable objects

Students from Imperial have used AI-powered cameras to identify recyclable objects, enabling them to be automatically sorted at waste facilities.

Students from Imperial College London have used AI-powered cameras to identify recyclable objects, and what they are made of, enabling them to be automatically sorted at waste facilities. 

According to research published in 2017, only 9% of the 8.3bn tonnes of virgin plastic produced worldwide has been recycled. Identifying items in waste streams is difficult as they are often highly damaged, can be half hidden and are mixed with other items.  

Replicating the power of human vision

Recycleye replicates the power of human vision to identify every item in entire waste streams, broken down by material, object, and even brand. By merging artificial intelligence and robotics, the team puts intelligence at the heart of the recycling process and transforms removal chains into supply chains. 

They are developing an ultra-low-cost sensor capable of classifying all items in waste streams by material type and brand. The logs generated will enable waste pickers, traders and facility managers to understand levels of contamination, and place an exact value on each tonne of recycled plastic. Brand-level detection will also extend producer responsibility for waste from their products.  

Play video

Autonomous sorting solution

Using low-cost robotics, Recycleye say their autonomous sorting solution will also be cheaper than current alternatives. At present the cost of sorting materials at plants is expensive – the industry uses multiple large, expensive sensors, which can miss some items. Using computer vision means Recycleye can use just one sensor for the whole waste plant, making it more cost effective.  

Peter Hedley, the company’s chief technology officer, said: “The human eye only sees visible light, but the power of our brains makes it the world’s best sensor. Unlike a machine we do not need to read a barcode to know the brand of an item, nor do we need laser spectroscopy or x-ray vision to know the material of an item. At Recycleye we are developing an AI that replicates this unique human ability.”

Recycleye, founded by PhD student Victor Dewulf and graduate Peter Hedley, have recently received £0.5 million in funding from Innovate UK to support their work.  

Co-founder Victor added: “Our motto is that waste doesn’t exist, it’s just materials in the wrong place. This project will accelerate the world’s transition to a circular economy and enable the merger of removal chains back into supply chains.”