This dataset contains part of the data used to generate the comparative analyses in our paper Efficient and Distributed Large-Scale 3D Map Registration using Tomographic Features. The associated code to run the experiments can be found on GitHub.
This README file was generated on 2024-09-12 by H. Utku Unlu The DOI for this dataset is https://doi.org/10.58153/m859g-t4p13
| Name | Role | ORCID | Institution | Email | | ------------ | ------ | ------------------------------------------------------------ | ----------- | ---------------------| | H. Utku Unlu | Author | 0000-0002-4368-5172 | NYU | utku@nyu.edu | | Anthony Tzes | PI | 0000-0003-3709-2810 | NYUAD | anthony.tzes@nyu.edu |
KITTI data is generated from KITTI benchmarka1-vdbmap raw data was collected on 2023-12-18, and processed to obtain the data used on 2024-06-13.KITTI dataset is distributed under the Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License, following the original dataset.
a1-vdbmap dataset is copyrighted by us and published under the Creative Commons Attribution-NonCommercial-Sharealike 4.0 License.
InteriorNet dataset is omitted due to licensing, but the raw data is accessible from the dataset website. Please contact the authors to obtain the processed InteriorNet data after agreeing to the relevant license.
The repository contains the maps generated from KITTI odometry benchmark in the folder kitti, and the large-scale indoor data generated by the authors, in the folder a1-vdbmap. The data is provided in a single compressed file named tomographic-map-matching-data.zip. For each individual map, the file name is generated using the template:
<category>-g<grid-size>-traj<idx>-n<noise-level>.pcd
where <category> is a1 for the large-scale indoor data, and one of kitti{00, 02, 05, 06, 07} for KITTI to denote the source data sequence, <grid-size> is the leaf size of the voxel grid filter applied to the maps, <idx> is the trajectory index, and <noise-level> is the magnitude of the noise applied during data generation, which applies only to InteriorNet data. Different trajectories that have the same category, grid size, and noise levels form the pairs to simulate a map matching task.
Each point cloud file has an associated .txt file, with -gtpose appended in the name string before the file extension, that stores the ground truth transformation in the same manner as the KITTI dataset.
kitti dataset trajectories are generated by splitting the odometry track of a particular sequence (00, 02, 05, 06, or 07) in half with respect to time, aggregating the point clouds after transforming them to the frame of reference of the first point cloud, and applying the voxel grid filter to the combined point cloud by the denoted size.
a1-vdbmap point clouds are generated over three separate trajectories followed by a ground robot, which are incorporated into a single map using VDBFusion algorithm, and extracting the vertice coordinates to generate the combined point cloud.
More details on the trajectory generation procedure can be found in our paper.
├── a1-vdbmap
│ ├── a1-g0.05-traj1-n0.00-gtpose.txt
│ ├── a1-g0.05-traj1-n0.00.pcd
│ ├── a1-g0.05-traj2-n0.00-gtpose.txt
│ ├── a1-g0.05-traj2-n0.00.pcd
│ ├── a1-g0.05-traj3-n0.00-gtpose.txt
│ └── a1-g0.05-traj3-n0.00.pcd
└── kitti
├── kitti00-g0.20-traj1-n0.00-gtpose.txt
├── kitti00-g0.20-traj1-n0.00.pcd
├── kitti00-g0.20-traj2-n0.00-gtpose.txt
├── kitti00-g0.20-traj2-n0.00.pcd
├── kitti00-g0.50-traj1-n0.00-gtpose.txt
├── kitti00-g0.50-traj1-n0.00.pcd
├── kitti00-g0.50-traj2-n0.00-gtpose.txt
├── kitti00-g0.50-traj2-n0.00.pcd
├── kitti02-g0.20-traj1-n0.00-gtpose.txt
├── kitti02-g0.20-traj1-n0.00.pcd
├── kitti02-g0.20-traj2-n0.00-gtpose.txt
├── kitti02-g0.20-traj2-n0.00.pcd
├── kitti02-g0.50-traj1-n0.00-gtpose.txt
├── kitti02-g0.50-traj1-n0.00.pcd
├── kitti02-g0.50-traj2-n0.00-gtpose.txt
├── kitti02-g0.50-traj2-n0.00.pcd
├── kitti05-g0.20-traj1-n0.00-gtpose.txt
├── kitti05-g0.20-traj1-n0.00.pcd
├── kitti05-g0.20-traj2-n0.00-gtpose.txt
├── kitti05-g0.20-traj2-n0.00.pcd
├── kitti05-g0.50-traj1-n0.00-gtpose.txt
├── kitti05-g0.50-traj1-n0.00.pcd
├── kitti05-g0.50-traj2-n0.00-gtpose.txt
├── kitti05-g0.50-traj2-n0.00.pcd
├── kitti06-g0.20-traj1-n0.00-gtpose.txt
├── kitti06-g0.20-traj1-n0.00.pcd
├── kitti06-g0.20-traj2-n0.00-gtpose.txt
├── kitti06-g0.20-traj2-n0.00.pcd
├── kitti06-g0.50-traj1-n0.00-gtpose.txt
├── kitti06-g0.50-traj1-n0.00.pcd
├── kitti06-g0.50-traj2-n0.00-gtpose.txt
├── kitti06-g0.50-traj2-n0.00.pcd
├── kitti07-g0.20-traj1-n0.00-gtpose.txt
├── kitti07-g0.20-traj1-n0.00.pcd
├── kitti07-g0.20-traj2-n0.00-gtpose.txt
├── kitti07-g0.20-traj2-n0.00.pcd
├── kitti07-g0.50-traj1-n0.00-gtpose.txt
├── kitti07-g0.50-traj1-n0.00.pcd
├── kitti07-g0.50-traj2-n0.00-gtpose.txt
└── kitti07-g0.50-traj2-n0.00.pcd
```