LiDAR (Light Detection and Ranging) systems are increasingly used from various mobile platforms (airborne, land, naval) for applications as varied as the inventory of road infrastructures, the study of coastal erosion phenomena, the calculation of the volume of material extracted from mines, etc. The future market for civil UAVs is booming in the field of terrain mapping. Their ease of use, precision and ability to navigate in complex environments make UAVs a tool of choice.
Automatic calibration and systematic error detection
This research and development project focuses on improving the performance of LiDAR measurement systems carried by mobile devices (ships, vehicles), which are increasingly used in the fields of coastal erosion monitoring, inspection of marine infrastructures, and coastline surveys. Our project targets the users of hybrid component systems, and aims at developing software tools to improve the performances through original calibration algorithms, and to provide numerical tools for objective evaluation of the performances of these systems, in order to objectively determine their limits of use according to the accuracy constraints imposed by certain applications.
Our project targets the users of hybrid component systems, and aims at developing software tools to improve the performances through original calibration algorithms, and to provide numerical tools for objective evaluation of the performances of these systems, in order to objectively determine their limits of use according to the precision constraints imposed by certain applications. The objective of this project is to develop automatic calibration methods that are as objective, precise and accurate as possible, in order to greatly reduce the level of uncertainty of LiDAR systems carried by mobiles (land vehicles, ships).
- To improve the existing methods of LiDAR system calibration, and to integrate them in an independent software.
- In the same spirit, develop performance validation methods, as well as an error propagation model to predict performance in terms of accuracy and measurement precision for a given application.
- Integrate technologies for automatic calibration and systematic error detection on autonomous platforms (autonomous vehicles, UAVs, asv, auv).
Mobile LIDAR system calibration and performance analysis
The approach adopted for the definition of calibration algorithms for LiDAR systems is based on mathematical models of the effect of systematic errors on geo-referenced points. These models are complex because they take into account all sources of error that may act on each point, but can nevertheless be expressed in a parametric form.
The originality and rigour of our approach also lies in the fact that the uncertainties related to the observations coming from the different sensors (LiDAR, INS, GNSS) are propagated in a georeferencing model. Each sensor is thus affected by a 3D uncertainty (which is in itself innovative compared to the methods used in current LiDAR data processing software), which can then be taken into account in the adjustment phase of the unknown parameters of the system. Indeed, the goal of our methods is to estimate the systematic part of the errors, and not the effect of statistical uncertainties on the LiDAR system observations.
The algorithms for calibrating the parameters of a LiDAR system can only work if the effects of certain errors are actually observable in the input data of the fitting methods. Therefore, it is necessary to select the appropriate data to ensure that systematic error parameters can be reversed. This will be done by developing observability criteria for each error taken independently in order to select the data allowing an unbiased estimation of the error parameters.
- Development of calibration methods in the laboratory (in lab) that do not require positioning ;
- Adaptation of calibration methods on natural terrain (in situ) to data from a land mobile LiDAR, and integration on a naval platform (in collaboration with the CCOM (USA));
- Validation of the methods in the context of land and naval mobile surveys;
- Definition of the sequence of application of each method according to the morphology of the terrain, the survey conditions, and the possible presence of other systematic errors.
In a second phase, these methods will be integrated into a software kernel for automatic calibration based on data from natural terrain, and data from laboratory tests allowing the user to :
- Define regions of analysis on overlapping areas of LiDAR data;
- Automatically extract the regions suitable for calibration of the various parameters of the LiDAR-IMU-GNSS system;
- Automatically adjust the error parameters;
- Correct the analyzed data set, and calculate performance indicators;
- Estimate a geo-referencing error model, incorporating the effect of all error parameters.
The scientific contribution of the project lies in the design of innovative calibration methods, operating in the laboratory or in situ, and the associated software. The technological contribution of the project is important because through this research and development work, we design automatic calibration and performance analysis methods that can be applied to a wide variety of systems. At the end of the project, the major expected result is a software package :
- Software for laboratory calibration of LiDAR parameters, without the need for positioning or absolute geo-referencing of LiDAR points. This software complements the laboratory calibration methods developed by Université Laval, and is intended to use a simpler and less expensive infrastructure than the methods currently developed by the CIDCO.
- Calibration software based on in situ data, with automatic selection of overlapping data and according to observability criteria of each calibration parameter.
- Software to verify the performance of LiDAR systems.
- Software to predict the total error propagated by a LiDAR system taking into account the residual calibration error statistics and the performance of each sensor (LiDAR, INS, and GNSS positioning).
The objective of the research project is to develop a software for calibration, performance analysis, and simulation of an error propagation model that can be applied to any LiDAR system based on hybrid components. This represents a significant advance in the field of mobile mapping by LiDAR. Indeed, many software packages are dedicated to LiDAR data processing, but few are dedicated to calibration and performance analysis. The innovative character of the project lies in the fact that we propose to develop a software independent of any hardware manufacturer (and therefore objective), using calculation methods that are more efficient than existing software.
At the scientific level, the calibration methods proposed in this project are rigorous methods based on solid mathematical foundations. This very promising approach must be valorized and generalized to the field of mobile mapping by LiDAR.
To carry out this project, CIDCO has brought together an important international partnership, which supports the project notably in terms of the provision of infrastructure, equipment and LiDAR platform in order to minimize costs. The project can thus focus on the design and development of calibration and performance analysis software, which will be the real added value of the project.