Development of a TLS Real Time Monitoring System for Landslides Ryan A. Kromer1,2*, Antonio Abellan1,2,3, Jean Hutchinson2 Matt Lato4, Marie-Aurelie Chanut5, Laurent Dubois5 & Michel Jaboyedoff1 1
Risk Analysis Group, University of Lausanne, Lausanne, Switzerland,
[email protected] 2 Geomechanics Group, Queen’s University, Kingston, Ontario,
[email protected] 3 Scott Polar Research Institute, Cambridge, UK 4 BGC Engineering, Ottawa, Canada 5 Groupe Risque Rocheux et Mouvements de Sols (RRMS), Cerema Centre-Est, France
Key words: LiDAR, TLS, Automatic Point Cloud Processing, Real Time Monitoring
Monitoring slope instabilities and landslides with remote sensing approaches forms an important part of many risk management strategies. Remote monitoring can be challenging given difficult site access, bulky monitoring equipment and low project budgets. Additionally, quick installation is sometimes required for a temporary monitoring campaign. A promising alternative to slope radar and GB-INSAR technologies is using a terrestrial laser scanner (TLS) as a light, portable, cost-effective temporary or permanent landslide monitoring solution. Such systems require automisation of the data collection and processing workflow given the vast amounts 3D data collected. In this study we address the development of both the hardware and software components of such a system. We built a wooden encasement to house an Optech Ilris long range TLS (see Figure 1) and anchored it to the first level of a concrete building overlooking the Séchilienne rockslide in the French Alps (HELMSTETTER & GARAMBOIS, 2010). Within the encasement, the TLS is mounted on a manual pan tilt allowing adjustment of the view angle. Data processing is conducted on site using a field laptop within the building. The laptop is connected to the local cellular network allowing for remote operation of the scanner and remote access to the processed results.
Figure 1: Field setup of the real time TLS monitoring system using Optech Ilris LR scanner.
The software component of the system consists of modules to operate the scanner, to manage and backup data, and automatically treat the data. The data processing module was developed in house using C++ with QT and the Point Cloud Library (RUSU & COUSINS, 2011) and is outlined in Figure (2). The first step of the algorithm is removal of unwanted points and a quality control (QC) step. Unwanted points are removed using a passthrough filter and the QC step consists of rejection of a point cloud if it does not contain a specified number of points, which is commonly due to poor atmospheric conditions or rainfall. The second step is registration of the point cloud to a reference through a registration pipeline consisting of an optional
initial alignment stage followed by an iterative fine alignment stage. The optional initial alignment consists of finding repeatable keypoints in the point cloud, defining descriptors based on the local keypoint point neighbourhoods and finding correspondences between features to perform an initial transformation. Refined alignment is conducted by iteratively transforming the point cloud, finding correspondences and using a rejector pipeline to discard poor correspondences until a convergence criterion is met. Change detection is conducted by calculating slope dependent change vectors and filtering noise using neighbours in space and time (KROMER ET AL., 2015).
Figure 2. Automatic TLS data acquisition and automatic data processing
Initial testing of the system reveals that the data treatment on the laptop is faster than the 10 KHz data acquisition rate, allowing near real time analysis and visualisation of change. This system can be a viable alternative to GB-InSAR monitoring given its relatively lower operating cost, ease of setup and high-resolution point representation of the terrain.
Acknowledgements: We would like to acknowledge the Centre for studies and Expertise on Risks, Environment, Mobility, and Urban and Country planning (CEREMA) for allowing installation of the TLS on their monitoring centre.
References HELMSTETTER, A. & GARAMBOIS, S., 2010. Seismic monitoring of Séchilienne rockslide (French Alps): Analysis of seismic signals and their correlation with rainfalls. Journal of Geophysical Research: Earth Surface, 115 (F3). KROMER, R.A., ABELLÁN, A., HUTCHINSON, D.J., LATO, M., EDWARDS, T. & JABOYEDOFF, M., 2015. A 4D Filtering and Calibration Technique for Small-Scale Point Cloud Change Detection with a Terrestrial Laser Scanner. Remote Sensing, 7: 13029-13052. RUSU, R.B. & COUSINS, S., 2011. 3D is here: Point Cloud Library (PCL). Robotics and Automation (ICRA), 2011 IEEE International Conference on, Shanghai: 1-4.