Viking Hill Dataset: A Lidar-Radar-Camera Dataset for Detection and Segmentation in Forest Scenes (ROS2 JAZZY VERSION)

The dataset captures a forested area that contains both fully grown and young trees, dense undergrowth, bumpy terrain and occasional rocks. Clearpath Husky robot was teleoperated through the area in May and June 2024, each time driving roughly one hour while recording its onboard sensors: LiDAR, high-resolution 4D radar, GNSS, RGB camera, IMU and internal odometry. Thanks to the GNSS receiver and lidar data, reference lidar point cloud map and thus absolute positioning of the robot during the experiment are available. Moreover, to support 4D radar research in detetection and classification of obstacles in similar environments, the dataset provides labels (3D cuboids) for several classes (e.g., tree trunk, tree canopy, bush, ...) and ROS2 tools that use these cuboids to label arbitrary point clouds in the dataset (lidar, radar, accumulated versions of those, or sub-maps).