Robotics in surface technology - with machine intelligence for autonomous production

Ever smaller batch sizes with increasingly individual customer requirements and complex set-up times: Düspohl from Schloß Holte-Stukenbrock was also faced with this challenge. The solution: The internationally active developer of profile wrapping machines as well as laminating and peripheral systems for the wood and plastics industry opted for end-to-end automation of the process chain. The aim was to minimize set-up times and implement the production of small batch sizes with individual decors efficiently and flexibly. Together with the Fraunhofer IEM, Düspohl developed the world's first fully automated profile wrapping system as part of the RoboWrap project launched in 2012. The machines are now being used by well-known profile manufacturers and the technology is being continuously developed further.

The automation of the wrapping process was achieved using a large number of intelligently programmed robots. The researchers at Fraunhofer IEM developed the software architecture, worked out the robot arrangement and specified the machine topology. Various questions had to be answered and requirements defined: How can the robots work with the necessary high forces and yet be compliant? How can they find their exact position? How can angular errors be avoided when applying the decor? And how can a smooth configuration be guaranteed?

The project team, which was awarded the OWL Innovation Prize for its work in 2012, developed a system that can produce new profiles fully automatically on the basis of artificial intelligence - without exceeding a set-up time of five minutes. While the positioning of the pressure rollers is currently still programmed by the machine operator, this step will also be automated in the future. As part of a joint AI marketplace project, Düspohl and Fraunhofer IEM are currently working on automating the system's set-up process and enabling fully autonomous production of any profile. To this end, Fraunhofer IEM is combining conventional modeling with machine learning and demonstrating how intelligent work systems can be designed in German SMEs in the future.