UAV-based LiDAR Mapping with Galileo-GPS PPP Processing and Cooperative Navigation

Towards a Multi-Pixel Time-of-Flight Indoor Navigation System for Nano-Drone Applications

“…While SLAM can be enabled at a low cost with relatively inexpensive and limited cameras, the process involves large data volumes and is often marred with limited mapping accuracy [34]. On the other side, LiDARs are significantly more precise for depth estimation and are commonly used for applications involving high-speed moving vehicles such as self-driving cars and drones [35]. LiDAR-based systems typically provide sparse samples organized into high-precision 2D or 3D point clouds.…”

“…LiDAR-based systems typically provide sparse samples organized into high-precision 2D or 3D point clouds. Even if they yield accurate mapping results when combined with SLAM, LiDARS are generally expensive and heavy, weighing a few hundred grams [35].…”

“…Many available tools already provide plug-and-play SLAM solutions that could be paired with other tasks such as sensor fusion, object tracking, path planning, and path following [36]. Although the mapping task seems to be a solved research problem in the literature, it relies on strong assumptions, such as memory availability of several gigabytes and powerful processor, e.g., the Intel i7 family [35], [37]. Moreover, carrying heavy and power-hungry 3D scanners, such as stereo cameras and LiDARs, is not considered a limitation for conventional robotic applications [35]- [37].…”

“…Although the mapping task seems to be a solved research problem in the literature, it relies on strong assumptions, such as memory availability of several gigabytes and powerful processor, e.g., the Intel i7 family [35], [37]. Moreover, carrying heavy and power-hungry 3D scanners, such as stereo cameras and LiDARs, is not considered a limitation for conventional robotic applications [35]- [37]. However, these assumptions do not hold for miniaturized and low-power robotic platforms, where the hardware cost is a concern, the payload is limited to a few tens of grams, and the computation power budget is limited to hundreds of mW [7], [11], [38].…”

“…Table I provides an overview of SoA mapping strategies in the UAV field, encompassing sensor types, mapping accuracy, and power consumption of the computing platforms. For example, Causa et al [35] proposed a scalable mapping strategy based on LiDAR and GNSS, utilizing a standard-size UAV weighing 3.6 kg with off-board processing. Shen et al [22] focused on onboard intelligence, utilizing a power-hungry Nvidia Xavier (30 W) and a VLP-16 LiDAR.…”