Discrepancies of LiDAR registrations to capture potential displacements of mechanically stabilized earth walls

This study considers terrestrial LiDAR to potentially capture undesirable displacements in mechanically stabilized earth walls (MSEWs), over time. In the USA, MSEWs have been used for almost 50 years and are becoming ubiquitous in bridge abutments. Time brings deterioration and the need to design and employ adequate inspection protocols for monitoring the integrity and deterioration rate of these structures. In that regard, the authors are closely working with the Georgia Department of Transportation (GDOT), USA. The current work aims to compare two scan-registration (scan-stitching) techniques to determine their relative accuracy. They are the target-based (TB) and the visual-aligned (VA) approaches. The relative spatial accuracy of their resulting virtual, 3D, point-cloud models, were compared against measurements completed in the field via a slow robotic total-station (RTS) instrument, with one-second angular accuracy, serving as the benchmarking device. The main goal of this study is to determine the minimum amplitudes of wall displacements that could be captured by the TB and VA approaches. The resulting georeferenced models showed that both techniques, TB and VA, presented almost the same accuracy when modeling MSEWs covering relatively small areas (0.23 and 0.38 ha). In a root-mean-square sense, the measurements obtained from the first analyzed bridge could capture most wall displacements larger than ~24 mm. Similarly, for the second bridge, both approaches could capture most wall displacements larger than ~11 mm. An additional multi-user analysis was completed to estimate the variability of the results due to users having different levels of experience with these techniques.