High-accuracy 3D mapping in GNSS-denied environments remains a major challenge for mobile robotic surveying, especially in harsh environments like underground spaces and adits, while the same applies to indoor environments where GNSS positioning is unavailable. This contribution presents a floating platform with a multi-sensor system (MSS) setup that combines an underwater LiDAR (ULi) system with a robotic-grade LiDAR for positioning and navigation purposes (Navi-LiDAR). In the specific use case, which is a partially flooded adit, the Simultaneous Localization and Mapping (SLAM) approach utilizing the Navi-LiDAR is applied to generate a spatially consistent 3D reconstruction of the environment under and above the waterline without GNSS.
In the proposed setup, the Navi-LiDAR is used to estimate the motion trajectory of the floating platform, while the ULi system performs high-resolution 3D data acquisition of the surrounding environment in air as well as of underwater structures in the adit. The two LiDAR systems are coupled through time synchronization, lever-arm and boresight calibration, and trajectory-based direct-georeferencing. Based on the SLAM-derived trajectory, each ULi system measurement is transformed into a common platform frame, allowing a complete 3D point cloud to be reconstructed even in environments where GNSS positioning is not possible.
The focus of this demonstration is on the interaction between navigation and mapping sensors: Navi-LiDAR and Inertial Measurement Unit (IMU) provide the spatial reference through SLAM, while the ULi system contributes dense geometric measurements for detailed 3D mapping. The resulting workflow supports applications such as underground water systems, flooded infrastructures, tunnels, reservoirs, and other challenging environments where conventional surveying methods are limited. The contribution highlights the system integration, georeferencing workflow, first mapping results, and key challenges related to trajectory accuracy, sensor synchronization, dual-LiDAR- and LiDAR-IMU-calibration, and motion-induced 3D point cloud errors.