Exploring the complete potential of mobile laser scanning raises the question of whether the obtained results are sufficiently accurate for deformation monitoring. This article focuses on the geometric quality of mobile laser scanning and proposes a method to enhance it to a level where it can be effectively employed for civil infrastructure monitoring. This research is motivated not by mere curiosity, but by an urgent necessity.
Civil infrastructure is a cornerstone of modern society, enabling transportation of goods and people across long distances. However, maintaining the safety and integrity of tunnels, bridges, noise barriers and retaining walls is a challenging task, particularly as infrastructure operators face increasing demands from growing traffic volumes and the vagaries of extreme weather. Failing to spot signs of deterioration can lead to severe consequences, such as soaring repair costs and even catastrophic structural failures.
Sadly, such a tragedy struck the Brenner Highway, a key north-south artery for both freight and passenger traffic across the European Alps. In 2012, a retaining wall adjacent to a toll station crumbled, sending tons of concrete cascading onto the highway (Figure 1). In a split second, a truck was buried and its driver tragically lost his life. This harrowing incident sparked a shift in Austria’s awareness regarding the hazard potential of geotechnical structures. There was an urgent need for more precise, efficient methods to detect defects in the tens of thousands of structures lining the country’s roads and railways.
Visual inspections and deformation monitoring
Guaranteeing the safety and reliability of infrastructure is paramount, and the assessment of its structural condition plays a crucial role in achieving this goal. Timely and well-informed decisions can be the difference between smooth transport of people and goods, or utter disaster. In many parts of the world, strict regulations apply to infrastructure maintenance, and cutting-edge technology is being deployed to gain deeper insights into the structural performance. While novel sensors and data analytics capabilities can help with building inspections and safety assessments, they cannot replace visual inspections as the ultimate basis for condition assessment.
Nonetheless, it is crucial to recognize the value of objective data. Various geotechnical or structural deficiencies may not be visible to the naked eye. The Eurocode 7 (EN 1997-1, 2004) spotlights the multiple failure modes that can lead to settlements, lateral displacements and tilting. That is where deformation monitoring becomes an indispensable tool, providing a standard means to observe behaviour and assess the current state of structural health. However, traditional techniques for deformation monitoring have their limitations. Measuring thousands of objects with total stations or installing tens of thousands of sensor nodes is impractical and can be incredibly time-consuming and resource-intensive.
Is mobile laser scanning the game changer?
Mobile laser scanning (MLS) has emerged as a cutting-edge technology for reality capture. Unlike static scanning, it allows for the acquisition of 3D point clouds of the surroundings while in motion, whether on foot, by air or by vehicle. The secret behind its success lies in direct georeferencing, which allows for the acquisition of 3D point clouds in the desired coordinate frame without the need for setting up and measuring ground control points. This is achieved by integrating sensor data from a variety of sources, such as GNSS, inertial measurement units, odometers, cameras and scanners. The quality of these individual components and the algorithms used to fuse the data play a critical role in the quality of the resulting point clouds.
For monitoring civil infrastructure along roads and railways, vehicle-based mobile Lidar systems are the ideal choice. They enable fast data collection while seamlessly integrating into the free-flowing traffic. Additionally, size and weight are less of a concern compared to systems based on uncrewed aerial vehicles (UAVs or ‘drones’), for instance.
In 2016, a team of interdisciplinary researchers from the Graz University of Technology in Austria began investigating the potential of a vehicle-based MLS system to identify potentially defective retaining walls. The aim was to collect data quickly and process it intelligently. The infrastructure would be scanned while driving by at high speed (Figure 2), and a reliable map of significant structural deformations would be extracted. Moreover, the research tackled the challenges faced when deploying the technology under practical boundary conditions. For example, infrastructure operators required a method that would work with commercially available systems. Data formats and interfaces were specified to guarantee data quality between different service providers and for years to come. Additionally, having well-defined guidelines would encourage more service providers to participate in tenders.
Evaluation of MLS accuracy
A successful collaboration with industry partners proved instrumental in creating a comprehensive database of MLS data. The project, consisting of eight measurement campaigns, involved the use of MLS systems from four different vendors to scan 24 support structures. The resulting dataset included hundreds of point clouds, each containing millions of points.
Analysis of this vast dataset highlighted some notable insights into the geometric accuracy of MLS systems. One significant finding was that the quality of the final data depends not only on the system used, but also on how it is operated. Factors such as the scanner’s orientation, measurement rate and calibration procedure – whether conducted on-site or in the factory – can have a significant impact on the accuracy of the data.
Moreover, the type of scenery being scanned can also affect the results. For example, the challenges posed by scanning alpine valleys – with retaining walls, bridges and tunnels – can be especially difficult. Consider the example of scanning an 18m-high retaining structure that stabilizes a slope next to an interstate road and a highway (as depicted in Figure 4). GNSS multipath and IMU drifts can cause cloud-to-cloud deviations of up to 10cm between two datasets acquired in short succession. It is worth noting that these deviations cannot be minimized by rigid-body transformation. The lack of software capable of dealing with such ‘MLS distortions’ may be the reason it has not yet proven itself as a reliable method for deformation monitoring.
Quality improvement in post-processing
However, there is a remedy. Systematic errors can be considered situation- or time-dependent. In other words, these errors are constant for a particular time frame. Most MLS software packages enable the export of point clouds with nanosecond-precise time information (e.g. in LAS/LAZ format). This information can divide the cloud into sections scanned within a concise period, which can then be aligned in terms of a rigid-body transformation. The potential for improvement is significant, as demonstrated in Figure 5. By post-processing MLS point clouds, one can eliminate systematic errors of several centimetres and align them with millimetre precision.
Putting the pieces together
MLS-based deformation monitoring has been missing a critical piece of the puzzle, which this method now provides. It is applicable at two key stages:
(1) Establishing a uniform coordinate system and deriving deformations
(2) Deriving empirical precision.
It is worth noting that the object from which the deformations are to be determined is excluded from the calculations in Stage 1, just like in traditional monitoring measurements with total stations. Semantic interpretation of point clouds can aid in separating the structure (object) from other elements such as the road surface and guardrails (reference). Two epochs are registered over these reference objects and the retaining structures are compared (Figure 6, top).
The precision from Stage 2 helps to differentiate between significant deformations and measurement noise. This leads to the creation of a binary map indicating areas of statistically significant differences, as seen in Figure 6 (middle). When compared to the results of total station surveys using 19 prisms (black dots in Figure 6, top), the results coincided within a range of +/-5mm (Figure 6, bottom).
The idea behind this concept, known as ‘rigorous deformation analysis’, is almost half a century old and was initially developed for tachymetric surveys. However, this study shows its transferability to mobile Lidar scanning data, which is intriguing.
Is it good enough?
As researchers and manufacturers strive to unlock the full potential of MLS, the question remains: Are the results accurate enough for deformation monitoring? The answer, as numerous studies have highlighted, depends on several factors. In optimal conditions with smooth object surfaces, a precision of +/-5mm can be achieved, even at high speeds of 80km/h. However, when dealing with rough objects or dense vegetation, the precision drops to an estimated +/-1cm, which is a more realistic assessment.
Despite these limitations, the overall quality of MLS data is deemed sufficient for the intended application, offering a promising tool for identifying critical infrastructure in need of attention. While static surveys will remain essential in critical scenarios, MLS has the potential to efficiently scan vast areas and pinpoint areas of concern, making it a valuable addition to the monitoring toolbox of infrastructure operators. As the world becomes more reliant on technology to optimize operations, it will be exciting to see how MLS continues to advance and play a role in maintaining and improving infrastructure.
Gabl & Pilch: Stützmauern Alt und Neu – Instandsetzung, Ertüchtigung und Neubau von nichtgeankerten Stützkonstruktionen. Verlässlichkeit auf allen Wegen, Kahlenberg, 18.06.2015.
Kalenjuk, S., Lienhart, W., & Rebhan, M. (2021). Processing of mobile laser scanning data for large‐scale deformation monitoring of anchored retaining structures along highways. Comput Aided Civ Inf. 2021: 1-17: https://doi.org/10.1111/mice.12656
Kalenjuk, S., & Lienhart, W. (2022). A Method for Efficient Quality Control and Enhancement of Mobile Laser Scanning Data. Remote Sensing, 14(4), . https://doi.org/10.3390/rs14040857
Kalenjuk, S., & Lienhart, W. (2023). Drive-by infrastructure monitoring: a workflow for rigorous deformation analysis of mobile laser scanning data. Structural Health Monitoring. Online-First: doi:10.1177/14759217231168997
EN 1997-1. (2004). Eurocode 7: Geotechnical design – Part 1: General rules. (Technical report, Authority: The European Union Per Regulation 305/2011, Directive 98/34/EC, Directive 2004/18/EC). Brussels.