Researchers at the Karlsruhe Institute of Technology (KIT) in the German city of Karlsruhe have developed a robotic system capable of estimating the failure in a malfunctioning product and then dismantling it while reducing the risk of damaging valuable parts. The system combines a predictive algorithm with robotic maneuvers that carry out the dismantling process, reviewing the results at each stage and updating the plan when unexpected behavior appears.
The importance of this approach is growing as industrial robots become more widespread in manufacturing sectors. According to the International Federation of Robotics, the number of industrial robots in use worldwide exceeds four million, while researchers expect the number to rise to more than 16 million by 2030 as industrial production expands. This raises the question of how to deal with these systems and products when they fail, rather than discarding them or replacing them entirely.
Dismantling That Adapts to the Failure
The KIT system relies on a CAD model of the malfunctioning product and each of its parts, enabling it to understand how the components move and detect deviations from expected behavior. It also uses a mathematical model to estimate the damage affecting the part. For example, a worn part may move a shorter distance than expected, while a loose screw may move farther, and the degrees of freedom of a deformed part may differ from those in its original design.
At the beginning of the process, the system forms an initial hypothesis about what may have happened inside the product, then tests that hypothesis by moving the components and monitoring their responses. If the result does not match the prediction, the system adjusts its approach. In the example presented by the researchers, the system initially tried to loosen a screw, but after simulating a stuck screw and observing that it remained in place, it switched to grinding away the surrounding material to remove the part.
The system also makes it possible to identify the parts that must be kept intact, allowing it to adjust the dismantling strategy according to their importance. This approach reduces the likelihood of carrying out many steps before discovering that a component is stuck or damaged in a way that requires a different method.
From Dismantling to the Circular Economy
Jan Baumgartner, one of the system’s designers, said that manufacturing a product using new parts is more straightforward because its steps are defined and do not involve expected deviations, whereas dismantling a malfunctioning product is unpredictable. He believes the system could help repair old devices instead of discarding them, reducing waste and supporting the circular economy.
Baumgartner’s future vision is to expand the system to include many robotic arms, each using specialized tools at a different stage of the dismantling process. This could take the form of a factory dedicated to dismantling a wide range of products, similar to industrial robot factories that manufacture cars, but with operations directed toward recovering components.
If expanding the system succeeds, it could become part of an automated process that extracts the failed component, replaces it, and then rebuilds the device. Baumgartner’s stated goal is to reach a cost that makes repairing electronic devices less expensive than producing new ones.
This research was presented at the 2026 International Conference on Robotics and Automation (ICRA) in Vienna. The material was updated on August 11, 2026, to clarify that the dismantling system is designed for products in general, not only robots.