Abstract
Manual assembly processes are often plagued with a lack of transparency while holding great potential for optimization. As preparing instructions takes a lot of time, assembly processes are often poorly documented, particularly for small volumes. This paper presents a general approach to automatically generate optimized process step sequences in manual assemblies from sensor data. The idea is to analyze the data and compare different assembly processes captured by the analysis. From this, a best practice can be derived and standard times can be defined for the individual process steps.