High-Resolution Inspection of Large-Scale Objects utilizing RoboCT

High-resolution X-ray computed tomography (CT) is an established non-destructive testing (NDT) method, but conventional systems are limited to components with diameters of a few hundred millimetres and moderate payloads. As a result, large and heavy parts—such as fully assembled battery systems for battery electric vehicles (BEVs)—are typically inspected using twodimensional radiography or destructive sampling, limiting defect detectability and inspection efficiency. To overcome these limitations, Fraunhofer EZRT has developed RoboCT, a twin-robotic CT system in which an X-ray source and detector are carried by two cooperating industrial robots. [1] This approach enables highly flexible, application-specific acquisition trajectories and region-of-interest (ROI) inspections on large-scale and complexly shaped objects. In this work, we present a scalable inspection workflow for large BEV battery systems that combines large-area two-dimensional radiography for fast screening with advanced three-dimensional imaging based on computed laminography (CL). In particular, we demonstrate Large-Field-Of-View Computed Laminography (LV-CL) using an algebraic reconstruction technique (ART), enabling volumetric reconstruction from limited-angle and sparse projection data. The LV-CL results are further refined by highresolution, ROI-focused multi-scan trajectories, achieving voxel sizes down to 30 µm in selected regions. Experimental results acquired on fully assembled and crash-damaged battery modules show that the proposed RoboCT-based workflow enables efficient identification of critical regions, improved separation of overlapping structures, and high-resolution inspection of safety-relevant features within meter-scale objects. These results demonstrate that RoboCT, in combination with modern high-energy microfocus X-ray sources and large-area detectors, provides a powerful and flexible NDT solution for quality assurance and failure analysis of large, safety-critical BEV battery systems. The presented results were obtained on a full-scale RoboCT system operated under realistic industrial constraints.