When a factory thinks for itself

This is not an advertising agency's slogan; it's something that Prof. Dr.-Ing. Gunther Reinhart of Munich Technical University proves every day: as the Director of the Institute for Machine Tools and Industrial Management (iwb) brings real machine tools into the virtual world of simulation, algorithms and neural networks.

"The machine tool is a particularly ideal choice for practising cognitive issues", explained Professor Reinhart. "Nowadays, you see, it runs highly complex operations." These involve firstly technology integration (e.g. laser machining and grinding in a single machine), which can be mastered and operated by staff only with the aid of computers, and secondly automatic handling operations that are susceptible to malfunctions and errors. Sensor networks can act as a preventive here, and computer programs (known as algorithms) can help in the event of malfunctions.

CoTeSys: detect, deduce, deliver

Artificial intelligence, sensor networks and algorithms: no doubt about it, the world of machine tools is turning virtual. For this purpose, research is ongoing within the framework of CoTeSys (Cognition for Technical Systems), an excellence cluster for developing cognitive technical systems, subsidised by the German Research Association. CoTeSys's remit includes incorporating artificial intelligence and fuzzy logic into the production machines. "We aim to use the tripartite requirement of "detect, deduce, deliver" to make machines more proactively autonomous than they used to be", says Reinhart. Nowadays, production lines are programmed to take some decisions; a network of sensors is now tasked with enabling the machines to obtain information about itself and its surroundings, so as to formulate possible solutions autonomously with the aid of knowledge databases, and translate them into appropriate action. To quote Reinhart: "For each solutional approach, we check whether it is able to meet the tripartite requirement:" "detect, deduce, deliver." This is not mere ivory-tower research: the "Cognitive Machine Shop (CogMaSh)" subproject has in Munich already created the first cognitive factory that thinks for itself - further facilities are under construction.

When workpieces can communicate

One aspect of CogMaSh is the "internet of things": for this purpose, workpieces are fitted with radio labels, known as RFID tags. For example, a component asks Machine Tool Number Two: “Can you drill a diameter of 50 millimetres on time, or shall I take myself to another machine?” The machine tool says “Yes”, the workpiece concerned uses RFID to book itself a “factory taxi”, and says to the driverless transport system: “Please take me to Machine Number Two.” After drilling has been completed, the workpiece stores information on its status, and the machining quality on its RFID tag.

The researchers at the iwb in Munich are also investigating the possibilities of virtual commissioning: they have succeeded in imaging a machine's physical characteristics, like density, centre of gravity, static and sliding friction coefficients, thus eliminating a defect of a CAD system's previous geometric data, not based on the fundamental laws of physics. "You then see a components hovering after a gripper has released it, although actually it ought to fall to the ground", explained the institute's Director Reinhart. "We're now incorporating proper physics intoNike Air Max 270