The International Journal of Robotics Research Study, Volume 43, Issue 8, Web Page 1114-1127, July 2024.
Over the last few years, innovations in Light Discovery and Ranging (LiDAR) modern technology have actually made 3D LiDAR sensing units a lot more portable, light-weight, and cost effective. This development has actually stimulated rate of interest in incorporating LiDAR with sensing units such as Inertial Dimension Devices (IMUs) and cams for Synchronised Localization and Mapping (BANG) study. Public datasets covering various situations, systems, and point of views are critical for multi-sensor combination bang researches, yet most concentrate on portable or vehicle-mounted gadgets with front or 360-degree sights. Information from airborne automobiles with downward-looking sights is limited, existing appropriate datasets typically include reduced elevations and are mainly restricted to tiny school settings. To load this space, we present the Multi-sensor Aerial Robots bang dataset (MARS-LVIG dataset), supplying special airborne downward-looking LiDAR-Visual-Inertial-GNSS information with point of views from elevations in between 80 m and 130 m. The dataset not just uses brand-new facets to examine and review existing bang formulas, yet additionally brings brand-new difficulties which can assist in r & d of advanced bang formulas. The MARS-LVIG dataset includes 21 series, gotten throughout varied large-area settings consisting of an aero-model landing strip, an island, a country community, and a valley. Within these series, the UAV has rates differing from 3 m/s to 12 m/s, a scanning location rising to 577,000 m2, and limit course size of 7.148 kilometres in a solitary trip. This dataset envelops information gathered by a light-weight, hardware-synchronized sensing unit bundle that consists of a solid-state 3D LiDAR, a global-shutter RGB video camera, IMUs, and a raw message receiver of the International Navigating Satellite System (GNSS). For formula examination, this dataset launches ground fact of both localization and mapping, which are gotten by on-board Real-time Kinematic (RTK) and DJI L1 (post-processed by its sustaining software application DJI Terra), specifically. The dataset can be downloaded and install from: https://mars.hku.hk/dataset.html.
发布者：Haotian Li，转转请注明出处：https://robotalks.cn/mars-lvig-dataset-a-multi-sensor-aerial-robots-slam-dataset-for-lidar-visual-inertial-gnss-fusion/
