Skaloud, JanBrun, Aurélien Arnaud2024-01-152024-01-152024-01-152023-07-13https://infoscience.epfl.ch/handle/20.500.14299/202941With the aim of improving trajectory estimation and point cloud georeferencing in the scope of Kinematic Laser Scanning (ALS), we recently proposed a novel procedure leveraging the automated extraction of reliable 3D point to point correspondences and their joint integration with raw inertial and GNSS observations in a bundle adjustment [1]. Here we put forward a more in depth analysis on the benefits of introducing these correspondences in the trajectory adjustment by considering new quality metrics and scenarios. Firstly, we highlight the improvements in term of accuracy and precision of the pose estimation and point cloud georeferencing under various scenarios such as GNSS outage and sparse correspondences. Secondly, we illustrate the benefits of introducing LiDAR correspondences in the scope of model’s parameters estimation. The system solved during the bundle adjustment contains millions of parameters and the delivered statistical analysis uncovers the impact of these new constrains in terms of improved parameters decorrelation and thus observability. Among those, particular attention is given to system mounting parameters, which put in light the capacity to deliver an accurate calibration simultaneously with the trajectory solving within a geometrical configuration that would be otherwise challenging to obtain. We close the discussion by raising open questions that motivates future investigations on the topic. [1] Brun, A., Cucci, D.A., Skaloud, J., 2022. Lidar point–to–point correspondences for rigorous registration of kinematic scanning in dynamic networks. ISPRS J. Photogrammetry Remote Sens. 189, 185–200. https://doi.org/10.1016/J. ISPRSJPRS.2022.04.027LidarGeoreferencingPoint cloud registrationUAVsPose-graph optimizationDynamic networkstopomappESOLABtext::conference output::conference presentation