The need for dynamic production and thus the ability to switch between mass production and batch size 1 is increasing. The end customer demands fast, punctual and correct delivery. These requirements are driving the intelligent networking of individual production processes and the integration of goods movements. To achieve this, data must not only be generated, but also used to optimize the value chain - a key challenge for the implementation of Industry 4.0. "At Hannover Messe 2019, we want to show our visitors how production and logistics can be networked in a meaningful way - and how value creation potential can be leveraged thanks to data transparency. In real time," explains Bernhard Müller, Senior Vice President Industry 4.0 at Sick.

To this end, sensor manufacturer Sick opened the doors of its smart production facility in Freiburg for the first time and brought them to the trade fair stand in Hanover. Visitors to the trade fair were able to follow a live stream of Automated Guided Carts (AGC) making their rounds in the '4.0 Now Factory', supplying production robots and transporting finished products away. The key figures for the site 600 km away can be called up on a dashboard. "All of our vehicles, components and production cells are connected to each other and deliver the data to a cloud. Production is scalable depending on the order situation and requirements. Automated and manual work take place side by side and combine the advantages of both variants for efficient production," explains Bernhard Müller.

The 'Bosch Production Performance Management' (PPM) of the '4.0 Now Factory' provides insights into the world of process and sensor data analysis in production: on the one hand, it shows how processes can be further optimized in terms of efficiency based on analyses. On the other hand, it shows how machine failures can be avoided based on predictive maintenance analyses and how service activities can be planned and implemented efficiently.

Sick also presented its first application based on deep learning algorithms in Hanover. Sick uses deep learning technology to specialize the functionality of sensors. The sensors learn to process information and thus acquire new functions. New processes also become possible when sensors supply, process and analyze data thanks to self-learning algorithms.

AI in real applications

For example, a sensor is trained to give an answer to a specific question using a large number of images. Based on this training, the sensor can then independently assign new, unfamiliar images to a result. "For example, we are currently working on a pilot project in the timber industry using deep learning. Our solution is based on a camera with deep learning functionality," explains Müller. To make optimum use of wood as a raw material, sawmills need to know what the conditions are in the log - where are the annual rings, where is the core? "We used deep learning to teach the camera how to make the best use of the wood. A task that previously could only be done by humans," adds Müller.