Comment on page

LiDAR SLAM Datasets

KITTI Autonomous Driving

  • Karlsruhe Institute of Technology (KIT), German, 2012
  • Provides a variety of modalities (vision, LiDAR, IMU, RTK, etc.) and application domains (odometry, stereo, depth estimation, semantics, object detection, etc.)
  • The ground truth in sequence 08 is not accurate at the beginning.
  • For odometry, the accuracy is evaluated by taking a sliding window of size 100, 200, ..., 800 frames, respectively, and computing the transformation difference between the first pose and the last pose in this window, and comparing it with the corresponding ground truth difference. The step size for the sliding window can be a fixed number such as 10 frames. The result is averaged over the entire trajectories and also different sliding window sizes.

NCLT Dataset

  • The University of Michigan North Campus Long-Term Vision and LIDAR Dataset, 2016
  • Mounted on a segway wheeled mobile robot.
  • Modalities: Camera, LiDAR, IMU, RTK GPS, Wheel Odometry

Newer College Dataset (NCD)

The Oxford RobotCar Dataset

The Oxford Radar RobotCar Dataset

MulRan Dataset for Urban Place Recognition

The Complex Urban Dataset

The HILTI SLAM Challenge Dataset 2022

The voxgraph Drone Dataset