PSI - Issue 78

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

1438

T5

Displacements acquired by the top level camera

L5 L4 L3 L2 L1

T5 is acquired by external camera Real displacements distribution: e.g., L3 disp. = T1-T3

shake table motion = T0-T1

L0

T1

shake table

input

ground camera

reaction mass

z

roof camera

horizontal and vertical shake table actuators

x

targets

Fig. 1. Basics of the proposed methodology during a shake table monodirectional input.

2.2. Three-dimensional configuration For three-dimensional monitoring, two internal cameras are placed on opposite roof corners to capture horizontal displacements along both global axes (x and y) and the shake table motion. This geometry maximizes measurement accuracy since the cameras are located at the maximum distance in plane, while improving sensitivity to possible floor rotations, assuming rigid-diaphragm behavior. Both roof cameras point downward to track multiple markers faced upward and distributed along the building height. Four external cameras complete the system, serving to ( i ) provide redundancy by recording roof displacements along each axis, and ( ii ) reduce noise in the internal camera data. Roof motion is captured through two dedicated targets—one per axis—positioned near each internal camera. The first application of the proposed vision-based methodology was carried out during Phase II of the shake table testing for the NHERI Converging Design Project, as shown by Barbosa et al. (2025). The test specimen was a six story mass timber building (Fig. 2a) with a structural layout consisting of beams and columns with pinned joints, designed to support gravitational loads. Lateral resistance was provided by four perimeter post-tensioned rocking walls, each detailed with a different lateral force-resisting system in the y- (North–South) direction, tested separately across the three project phases. In Phase II, the lateral force-resisting system comprised Buckling-Restrained Braces (BRBs). The goal of applying the vision-based methodology in this phase was to measure building displacements along the y-direction using only two video cameras—one internal and one external. The internal “roof camera” was mounted at the North–East corner of the roof (L7), directed downward to capture multiple visual targets along the building’s height. The external “ground camera” was positioned approximately 26 m from the East façade at ground level (L0), providing an independent measurement of the roof displacement for redundancy and noise compensation. The Field Of View (FOV) of each camera was defined considering the following factors: ( i ) specifications of the optical hardware; ( ii ) inter-story heights of the building, corresponding to the distances from the sensor; ( iii ) 3. First application to a full-scale shake table testing 3.1. Testbed building and design of the vision-based system