The trend towards modular production systems is increasing due to the complexity of the end products. You go one step further and build customized solutions. What requirements do you face?
David Rosas Wolf: In most cases, the increasing complexity must also fulfil a high level of future flexibility. In order to achieve sustainability here, we rely on modular designs. On the one hand, the use of modular designs achieves a standardization of the individual modules and assemblies, which is reflected in lower production costs and flatter spare parts logistics.
For the end customer, however, there are also further synergies. Technologies, which they use today for various products, can be adapted or replaced for future requirements without any problems. Modular designs allow the relevant assemblies to be upgraded or replaced. This ensures lower investment costs for the future, as the foundation can remain in place.
In view of the complexity mentioned, the requirements for control and inspection systems are also increasing. How do you meet these challenges?
David Rosas Wolf: Here, too, we take the path of modularity. We apply this to both the hardware and the software architecture. We have focused on human-machine communication and designed the system to be as flexible as possible through the use of individual templates. The operator can thus perform the most complex and diverse tasks under the familiar operating environment, without any additional training effort. Thus, the system is also infinitely adaptable for future tasks.
User-friendliness, automation, and artificial intelligence are currently highly discussed topics. What significance does this have for you and how are you adapting to it?
David Rosas Wolf: We’ve already experienced very positive experience with the use of AI in the inspection of labels for years. Particularly with the increasing complexity of products, AI systems are proving to be clearly superior to conventional systems in terms of robustness of the recognition rate and avoidance of over-detection. By using AI and “deep learning”, the operator learns such a system directly by means of their experience.
The assessment of whether a product is still good or not, no system can do this as well as an operator who has been doing this job for years. It is precisely this experience that can be transferred to the system. The result is that after a certain learning phase the system automates the process with almost the same precision. Due to the modular design, the training status can be saved and reloaded at any time in case of repetition. An inspection job thus learns with every repeat job and becomes more and more clever. Even as an engineer, it sometimes becomes mysterious how perfectly these systems then work under the most adverse conditions.
At present, the widespread use of AI is limited to subareas due to the still insufficient computing capacity. But the future will certainly remedy this.