SIEMENS

What’s the biggest challenge associated with tomorrow’s factories?

Zäh: Industry faces a constant dilemma between productivity and flexibility. On the one hand, companies want to manufacture as efficiently as possible, but on the other hand they are trying to accommodate very individualized demands. However, in order to be flexible you also need to be able to retool so that you can switch production from one product variant to another. Conventional automation is no longer enough if you want to achieve the most extreme form of flexibility, which is mass-customized manufacturing – an individualized mass product. Conventional automation is productive, but it’s also relatively inflexible. Reconciling this contradiction is therefore the biggest challenge at the moment.

What concepts can be used to reconcile these conflicting objectives?

Zäh: It can’t be done with conventional automation. What we need are cognitive production systems, and the most flexible system we have consists of humans and their skills. For example, we are able to master the so-called PCA loop, where PCA stands for Perception, Cognition, and Action. We humans can perceive things, process them cognitively, and then derive a course of action. The production systems of tomorrow must therefore include technologies with the ability to process things cognitively – in other words, they should be able to determine the meaning of what they perceive and then come up with an appropriate course of action.

How would they do that?

Zäh: Many of today’s factories already use robots with sensors, and these robots need to be further developed in a way that will enable them to react to information from their surroundings. They have to be able to respond to the changes in a given situation. However, this is not yet possible. For example, today’s robots don’t know what to do if two components don’t fit together or have too much play between them. Another key goal is to have people and robots working in the same spaces so that the robots can “read” human behavior in order to determine which actions are appropriate. We also need to make sure robots don’t present a danger to humans in shared workstations. Robot and human workstations are still strictly separated. Our vision is for robots to become colleagues for humans. At the moment, they are productive, while people are flexible; we want to combine the best of both worlds.

How do you turn a machine into an employee?

Zäh: There are several ways to give robots cognitive abilities. For example, you can develop a database that contains all possible instructions for action, as is already the case at the wikiHow Internet portal. Our computer scientists have developed algorithms that convert text from wikiHow into robot actions. The drawback here is that humans have to define all the eventualities in advance. An alternative approach is to develop robots with the ability to learn. Such robots might observe humans and learn to recognize their movement patterns, behaviors, and gestures. They would also need to process the information they collect in order to be able to mimic what we do. Learning robots could perform many services in an ideal manner because they would have been specifically programmed to focus on and imitate their human mentors. This is all still a long way off, but it would be perfect for resolving the conflict between productivity and flexibility.

Which approaches are being studied for creating human-robot workstations – tearing down the wall, so to speak?

Zäh: There are many variations, including foot mats that activate a circuit when a person steps on them. You can also combine a number of mats and thus resolve a movement pattern into several sectors. This would allow a robot to know where people are located, how they got there, and where they’re going next. Another approach involves laser scanners that probe the area above an assembly table – you can also use cameras for this, of course. In any case, a robot would then be in a position to perceive movements and reduce the speed it works at when it gets close to people so as to avoid scaring or injuring them. The idea is to have the robot do the strenuous work, for example by carrying heavy objects or placing tools where they’re needed. People, in turn, should carry out production operations that require fine motor skills or a high level of cognitive ability. People and robots would then be able to work together in the most efficient manner possible in the factory of the future – and productivity and flexibility would increase at an equal rate. This type of human-machine cooperation would also be very desirable from the perspective of our aging society.