Facial recognition technology normally used to track people has been redeployed by a London start-up to cut the volume of recyclables sent to landfill.
Waterloo-based Greyparrot’s system spots different types of waste on a conveyor belt and uses robot arms to scan and drop them into the correct bin.
The system uses artificial intelligence and cameras to identify variations in aluminium, paper, cardboard and newspapers in a fraction of a second.
The start-up says their software is more efficient than human eyes at waste transfer plants, and is more capable of finding the purest plastics needed for recycling. “Rubbish recognition” means the arm can identify different plastic items such as straws, a cola cap and milk bottle. It assigns them a plastic purity rating when they are sorted.
It is hoped the technology will reduce the recyclable waste that can slip through inefficient plants. The company also says it automates “an unpleasant manual task”.
The technology is being trialled in Britain and South Korea — where cameras are placed on conveyor belts at waste transfer stations — with the hope that London councils will use it in the future.
Mikela Druckman, the firm’s co-founder, said: “For the sorting facilities to be able to sell material to be recycled it needs to be of very high quality, otherwise it will be rejected, nobody will buy it and it won’t be recycled. Plastic bottles need 99.99 per cent purity to be recycled. We’re providing a monitoring system, placing cameras at the end of lines and estimating the level of purity that’s there.
“It’s very much the same core technology you would use for face recognition or self-driving cars.”
Richard Kirkman, chief technology and innovation officer at London waste giant Veolia, said one of the biggest challenges for computer vision is recognising crumpled plastics.
Greyparrot is also piloting “smart bins” in airlines and hotels so they can more efficiently sort recyclables.
