Harri Valpola spent the last 20 years teaching computers to mimic how the human brain processes information. Now, as cofounder and principal scientist of Helsinki-based ZenRobotics, he is using machine learning to help robots to sort valuable raw materials from construction and demolition waste.
Built with off-the-shelf components, the ZenRobotics Recycler (ZRR) uses a camera and sensors -- including near-infrared spectrometers, 3D laser scanners and metal detectors -- to identify waste objects. A computer program categorises items to be kept or discarded. "The robot is a standard industrial model: the magic is between the sensorsand the controller," says Valpola (pictured). His firm makes two models -- the €480,000 (£400,000) ZRR Fast Picker and the
€580,000 (£490,000) Heavy Picker, both capable of sorting wood, metal and stone. As the software is taught to recognise new materials, customers can easily upgrade.
In September 2012, ZenRobotics announced it had raised €13 million (£11 million) from investors to market the ZRR globally; in December, SITA Finland (part of SUEZ Environment, Europe's largest environmental services company) bought two units. The robots are taking over the world -- of recycling.
This article was originally published by WIRED UK
