New technologies such as machine learning, large language models, automation through intelligent robotics, quantum computing, embedded data acquisition and digital pipelines are leading to drastic changes in the generation of scientific knowledge. We need to harness these technologies to tackle the challenges of this decade such as climate change or resource scarcity. The talk will focus on an automated data-centred approach to knowledge discovery based on fatigue crack growth experiments. Robots are used to track the crack tip of a fatigue crack during an experiment for high resolution digital image data (DIC). These data sets are automatically analysed by a machine learning model to detect the crack tip and then the actual crack tip loads are evaluated. EBSD (Electron Backscatter Deflection) maps of the material and 3D scans of the fracture surfaces after the test provide information on how the crack has propagated in the microstructure. The consolidation of all these different data sources into a common knowledge graph enables cause-and-effect relationships to be recognised automatically. The aim is to shorten development times and bring new products to market more quickly.