Challenging SLAM dataset from Livox MID360 lidar mounted on K9 rescue dog

This dataset is collected during Search and Rescue training. Data collection tool was mounted onto K9 performing victim search mission. It is composed of two LiDARs LIVOX MID360 recording all data during task execution. The scenario is composed of two main challenges: traverse via forest (~300m), bunker inspection / search (~70m). The challenge is the K9 motion, and surrounding environment composed of trees and tight corriddors thus recent LIO (LiDAR Inertial Odometry) algorithms can be benchmarked in realistic conditions. This data set provides realistic recording from real-world application in search and rescue mission. Dataset contains multiple point cloud (.laz), IMU (.csv), GNSS (.gnss) and calibration files (.sn and .json) that create trajectory traversed by search and rescue canine. Compressed continousScanning_0078.zip directory contains multiple sets of (apoint cloud, IMU, GNSS and calibration) files: .laz - binary point cloud (compressed LAS format), .sn - serial numbers of lidars used co collect point cloud at given trajectory point, .GNSS - list of GPS coordinates collected in proximity of the place that given point cloud was collected, .csv - list of IMU data points collected in proximity of the place that given point cloud was collected, .json - file with extrinsic calibration between coordinate frames of 2 lidars that were used to capture dataset.