Assessment of accuracy and performance of terrestrial laser scanner in monitoring of retaining walls

Abstract

Retaining walls are a critical infrastructure of transportation networks and the monitoring of their condition is crucial for the efficient and reliable maintenance of the network. The condition of retaining walls is frequently assessed using qualitative criteria and visual inspection, which are susceptible to human-bias and errors. To improve the management of these structures, reducing the probability of failure and the maintenance costs, it is critical to develop more efficient, reliable and quantitative monitoring approaches for these structures. The current study aims to evaluate the performance of Terrestrial Laser Scanner (TLS) in deformation monitoring of retaining walls, based on the analysis of single scans (without registering the point clouds to build 3D models). The evaluation was based on a controlled experiment, where a wooden frame (1.5m x 1m) was used to simulate deformation scenarios for retaining walls, with an amplitude between 2 to 16 mm. A Leica RTC360 scanner was used to scan the wooden frame from distances varying between 10 to 27 m and angles varying between 0° to 20°. Five methods were applied to analyse the laser-scanner data and estimate the displacement: a target-based approach and four cloud-based approaches including the Cloud-to-Cloud (C2C), the Cloud-to-Mesh (C2M), the Multiscale-Model-to-Model-Cloud-Comparison (M3C2), and an alternative cloud-based method where the mean average of the point-cloud was used to estimate the displacement in the axis of the deformation. A Robotic Total Station Leica TS30 was also used to measure the deformation of the wooden frame and provide the ground truth values of the introduced deformation for each scenario. The results showed that the RTC360 had an accuracy of 1.3 mm with a confidence level of 95%.