Reinforcement learning is an artificial intelligence process that works in much the same way as most people learn to ride a bicycle - without knowledge of basic physics through trial and error.

"After about 72 hours of training with the Digital Twin (on a standard computer; about 24 hours on computer clusters in the cloud), the AI is ready to control the real plant. That is definitely much faster than when humans develop these control algorithms," says Bischoff. With reinforcement learning, the AI has developed a solution strategy in which all the chocolate bars on the front conveyor belts are transported on as quickly as possible and the exact speed is only controlled on the last conveyor belt - is interestingly quite different from that of a conventional control system.

"The chocolate bars are placed at random intervals on the first feed belt. Our AI controller now slows down or speeds up three other conveyor belts connected in series, ensuring that the chocolate is correctly positioned on the last belt," says Martin Bischoff, an expert in digital system integration from Siemens' Technology research department. "The control algorithm for this is definitely a tricky programming task - if you don't believe it: just try it yourself. We have now used reinforcement learning to train artificial intelligence to take over this control."