Human-centric artificial intelligence architecture for industry 5.0 applications

Figures & data

Figure 1. Intersection of architecture value-based principles, and architecture building blocks addressing them.

Three intersected circles describing the value-based principles for Industry 5.0 and architecture aspects realised in their intersection.

Figure 2. Proposed architecture contextualised within the BDVA reference architecture components.

Architecture diagram showing architecture modules and BDVA concerns.

Figure 3. The proposed architecture modules, a storage layer, and their interactions. In addition, we distinguish (a) the physical and digital worlds, (b) manufacturing platforms, (c) artificial intelligence systems, and (d) digital twin capabilities.

Architecture diagram showing the interaction between the proposed architecture modules.

Figure 4. Samples of three types of images: (a) good (no defect), (b) double-print (defect), and (c) interrupted print (defect).

Sample images regarding defective and non-defective manufactured parts for the Philips use case.

Figure 5. Sample screen for the manual revision process. We provide the operator an image of a non-defective part, the image of the component being inspected, and the hints regarding where we do expect the error can be. The images correspond to cases where the hints were created with (a) GradCAM, (b) DRAEM, and (c) the most similar labelled image.

Images showing different defect hinting approaches.

Figure 6. The sensing prototype for worker's activity recognition worn on the wrist.

Image of a prototype device for sensing worker's activity.

Data availability statement

Data not available due to restrictions.