Robots can’t mill metal? 

When the milling robot lowers its arm onto the metal plate and gets going, the chips fly off in all directions. Then, a few minutes later, there is a relief in the plate, a miniature landscape of terraced slopes of different shapes that are connected via channels. The mini-landscape is essential proof that robots can do things that no one thought they could do for a long time.

"When robots mill metal, they are subjected to enormous forces," says Birgit Obst. She says it has been difficult for robots to maintain their trajectory very precisely or to repeat such tasks even when different forces are acting on them. "They couldn't compensate for the forces in real-time." Obst works as a technology driver at Siemens Technology in Munich. Together with Dirk Hartmann, a senior scientist at Technology, and Florian Schnös, she has found a way for robots to perform such tasks. Schnös was a doctoral student at the Technical University of Munich (TUM) at the time and now works for KUKA AG.

Using simulation, the three scientists were able to show that robots can take on such tasks - as precisely as the massive machine tools that previously had the unique selling point for such tasks. They are being honored as Siemens "Inventors of the Year 2021" in the Open Innovation category.

"Robots are an important building block of Industry 4.0," says Dirk Hartmann, "but they couldn't do a few tasks that are crucial in the world of industrial manufacturing. " Not only when milling metal, but also in many other processes, strong forces were acting, and it was difficult for robots to very accurately maintain their trajectory and repeat tasks precisely. "We humans do this all day long," adds Florian Schnös. "We compensate for forces acting on us - usually without giving it much thought."

Robots need factual information to do this - and at the beginning of their project, they often heard that they couldn't mill "with a robot like that," Hartmann recalls. "We were able to do it with simulation," he says, and his joy is evident. The scientists used physical knowledge and numerical algorithms to do it. "Numerical algorithms can predict the forces that arise during the milling process," says Birgit Obst. "This means: using a component that is yet to be produced, these algorithms also calculate along the path that is milled the forces that can be expected at any time, which would act on the milling head, cause a displacement, and can now be compensated. "

In addition, the team was able to use data from the machining process, such as the torque on the spindle. By comparing the measured data with the predicted torques from the simulation, they can see if there are any additional deviations of the reality from the plan. "And then we can adjust that in real time," Obst says.

They worked closely with colleagues from Siemens Digital Industries for the project. This allowed them to use the domain knowledge and the already developed products and interfaces, especially Sinumerik - Run MyRobot /Direct Control and Run MyCC /ROCO (RObot COmpensation). Run MyRobot is a solution for the easy commissioning and optimal setting of a robot with Sinumerik. Based on an algorithm, ROCO considers the physics of the robot and increases its precision. “Our simulation delivers additional input for this module”, says Birgit Obst, “considering the forces that the milling process puts on the robot.” Run MyCC/ ROCO compensates for these forces and pushes the robot back on its original path. “We pushed our limits in the virtual world to make the production more efficient in the real world.” The prototypes have already been implemented. They can map the entire process chain end-to-end and do so with Siemens tools already established at the customer’s site, such as NX CAM and SINUMERIK.

The three project partners are pleased to offer customers real added value - simply by updating the software. Dirk Hartmann says: "I think this is how 'Software-as-a-Service' should ideally work for industry - similar to what we know from the entertainment industry: You bought a product, and a simple software update makes the product - in our case, industrial manufacturing - better."