Human Factors in Cyber-Physical Production
Making industrial data legible to the people who have to act on it, and putting the human back into cyber-physical production.
Production research has a long habit of modelling machines carefully and people casually. The result is a generation of cyber-physical systems that are technically sound and organisationally brittle, because the human in the loop was specified as a constant rather than studied as a variable.
My contribution to the Internet of Production ran along that seam. I led the work package on data-to-knowledge and knowledge-to-data pipelines, which is the unglamorous but decisive problem of turning the enormous exhaust of a modern factory into something a person or an algorithm can actually use. The related idea of the digital shadow — a purpose-built, reduced representation of a process rather than a full digital twin — extends naturally to people, and the human digital shadow is an attempt to model users and usage with the same seriousness the field already applies to machines.
Concrete strands included decision support for multi-level production systems, process recommenders for fibre-reinforced polymer and textile production, cross-company cooperation where trust has to survive an information asymmetry, and privacy in the smart factory, where the data being collected is increasingly about employees rather than about parts. The World Wide Lab work asked what becomes possible when production sites share pipelines rather than merely publish results.
The framing paper I return to most often is the computer science perspective on digital transformation in production, which argues that the interesting problems in this domain are not manufacturing problems with a software component but socio-technical problems throughout.
