PSI - Issue 78

Laura Gioiella et al. / Procedia Structural Integrity 78 (2026) 1436–1442

1441

The peak positive and negative values of the platen displacement are summarized in Table 1 to evaluate the precision of the vision-based measurements with respect to the shake table controller, using the Root Mean Square Error (RMSE) as the primary metric. The Normalized RMSE (RMSEN) is also reported, expressing the error as a percentage. The RMSE and RMSEN values are 3.21 mm and 8.09% in the y-direction, and 2.89 mm and 10.08% in the x direction. When considering only the percentage errors at the maximum positive (Max Error) and negative (Min Error) displacement peaks, the errors notably decrease to –5.09% and –2.30% in the y-direction, and 2.12% and 1.66% in the x-direction. It should be noted that the platen motion reported by the controller refers to the shake-table Center of Mass (CoM), whereas the roof camera measurements are referenced to the North-East (NE) corner of the structure. Therefore, slight discrepancies between the two datasets may be attributed to building rotations during testing. Table 1. Statistics of the shake table motion acquired by roof camera and controller. Direction Source Max [mm] min [mm] RMSE [mm] RMSEN [%] Max Error [%] min Error [%] y roof cam. 127.64 -169.39 3.21 8.09 -5.09 -2.30 controller 134.49 -173.37 x roof cam. 87.73 -174.98 2.86 10.08 2.12 1.66 controller 85.91 -172.12 4.Conclusions The experimental results confirm the effectiveness and practicality of the proposed vision-based methodology for displacement monitoring in shake table testing. The main findings can be summarized as follows:  The approach leverages computer-vision to measure structural displacements with high accuracy, achieving results closely matching those provided by the shake table controller.  The combined use of internal and external cameras effectively compensates for spurious displacements caused by vibrations affecting the internal camera during motion playback, thereby improving measurement reliability.  The method is highly efficient, requiring only a small number of cameras compared to the large arrays of contact sensors typically used in testing, significantly reducing setup time and operational complexity.  The system relies on cost-effective hardware (industrial cameras, lenses, artificial targets, and supports) combined with efficient video-processing software, enabling the potential for real-time extraction of displacement time histories for multiple points within each camera’s field of view. Overall, the methodology offers a robust, accurate, and scalable alternative to traditional displacement measurement techniques, with clear benefits in terms of cost, setup efficiency, and adaptability to various structural testing scenarios. Acknowledgements The Authors wish to thank Professor Joel P. Conte at the University of California San Diego (UCSD) and the staff of the UCSD LHPOST for their support during testing. The Authors of the University of Camerino acknowledge the financial support from the European Union - NextGenerationEU – "Piano Nazionale di Ripresa e Resilienza, Missione 4 Istruzione e Ricerca - Componente 2 dalla ricerca all'impresa - Investi-mento 1.5, ECS_00000041 VITALITY - Innovation, digitalization and sustainability for the diffused economy in Central Italy" and from the Italian Government “Piano Nazionale di Ripresa e Resilienza - Fondo Complementare Programma unitario di intervento per le aree del terremoto del 2009 e 2016 Misura B - Sub-misura B.4 Cen-tri di ricerca per l’innovazione Centro internazionale per la ricerca sulle Scienze e Tecniche dalla RICostruzione fisica, economica e sociale STRIC and STRIC+”. The Authors would also like to extend thanks to the graduate students Tanner Field and Steven Kontra from Oregon State University as well as Ludovica Pieroni from Uni-versity of College London for their help with the installation of instrumentation and testing. The shake-table testing and construction for the specimen for this project was executed under the National Science Foundation awards #2120683, #2120692, and