The valuable materials are already in our products. Getting them out is the hard part

Using fewer new raw materials means recovering the ones already in circulation, but many end-of-life products hold valuable components that are hard to extract intact. Danfoss faces this with its motor variable frequency drives. When they come back, the valuable parts sit deep inside, about 50 steps into the product. Shredding would destroy them, so a person currently has to take each unit apart by hand, step by step. Wilbert Peter Empleo, a PhD student at the University of Southern Denmark working in the MOTUS project, says that approach is not scalable.

The pressure to fix this is growing. Electronic waste increases every year, and most of the valuable metals and components in it are never recovered. Europe wants to depend less on imported raw materials and has set targets for recycling more of what it already has. New EU ecodesign rules will also increasingly expect products to be repairable and easy to take apart. More products will therefore come back, and more will need careful disassembly rather than shredding.

Robots look like the obvious answer, but disassembly consists of many small, varied tasks, and robots have been costly to teach. The end-to-end AI models behind modern robots turn a camera image directly into movement, and they usually need 30 to 100 demonstrations for each task. Empleo found a method that needs only one. The robot watches a task once and can repeat it even if the part has been moved or turned.

He demonstrated this live at Danfoss, where the company's own staff guided the robot with a 3D mouse while taking the product apart, and the robot learned from them. If each step only has to be shown once, far more products become worth disassembling.

MOTUS is a DIREC project involving the University of Southern Denmark, IT University of Copenhagen, Danfoss and Universal Robots. Empleo says working with the partners has improved his research and pointed it at a problem that matters.