To reduce resource consumption, used products should be transitioned into current product generations within a circular production pattern. The interdisciplinary special research project Circular Factory aims to realize the vision of the eternal product.

Various measuring and testing facilities are available to provide information and data. The main goal of this work is to fuse this data to make informed decisions regarding the further use of used products.

In this work, digital twins of the object instances will be introduced, providing information about various types of uncertainty and representing all quality-relevant attributes.

The challenges lie both in the variability in the condition of used products and in the heterogeneous information sources with different references and interpretations of uncertainties. The implementation of this information fusion requires the transformation of information into a probabilistic representation to perform efficient Bayesian fusion based on instance-specific tolerance schemes.