Identifying predictive EEG features for cognitive overload detection in assembly workers in Industry 4.0

Industry 4.0 will be characterized by far-reaching production automation because of recent advancements in robotics and artificial intelligence. As a consequence, a lot of simple, repetitive assembly tasks will no longer be performed by factory workers, but by machines. However, at the same time, consumers demand more and more personalized products, increasing the need for human assembly workers who can adapt quickly to new and more complex assembly procedures. This need for adaptation is most likely to increase the cognitive workload and potentially overload assembly workers that are used to traditional assembly work tasks. Several studies have tried to identify this cognitive overload in the EEG signal, but many failed because of poor experimental measurement procedures, bad data quality and low sample sizes. In this paper, we therefore designed a highly controlled lab experiment to collect EEG data of a large number of participants (N=46) performing an assembly task under various levels of cognitive load (low, high, overload). This systematic approach allowed us to study which EEG features are particularly useful and valid for cognitive overload assessment in the context of assembly work.