Cooperative Robotic Exploration of a Planetary Skylight Surface and Lava Cave - Datasets

The dataset contains the logs used to produce the results described in the publication Cooperative Robotic Exploration of a Planetary Skylight Surface and Lava Cave. Raúl Domínguez et. al. 2025

Cooperative Surface Exploration

- CoRob_MP1_results.xlsx: Includes the log produced at the commanding station during the Mission Phase 1. It has been used to produce the results evaluation of the MP1.

- cmap.ply: Resulting map of the MP1.

- ground_truth_transformed_and_downsampled.ply: Ground truth map used for the evaluation of the cooperative map accuracy.

Ground Truth Rover Logs

The dataset contains the samples used to generate the map provided as ground truth for the cave in the publication Cooperative Robotic Exploration of a Planetary Skylight Surface and Lava Cave. Raúl Domínguez et. al. 2025

The dataset has three parts. Between each of the parts, the data capture had to be interrupted. After each interruption, the position of the rover is not exactly the same as before the interruption. For that reason, it has been quite challenging to generate a full reconstruction using the three parts one after the other. In fact, the last one of the logs has not been filtered, since it was not possible to combine the different parts in a single SLAM reconstruction, the last part was not even pre-processed.

Each log contains:- depthmaps, the raw LiDAR data from the Velodyne 32. Format: tiff.- filtered_cloud, the pre-processed LiDAR data from the Velodyne 32. Format: ply.- joint_states, the motor position values. Unfortunately the back axis passive joint is not included. Format: json.- orientation_samples, the orientation as provided by the IMU sensor. Format: json.

- asguard_v4.urdf: In addition to the datasets, a geometrical robot model is provided which might be needed for environment reconstruction and pose estimation algorithms. Format: URDF.

Folders contents

├── 20211117-1112│   ├── depth│   │   └── depth_1637143958347198│   ├── filtered_cloud│   │   └── cloud_1637143958347198│   ├── joint_states│   │   └── joints_state_1637143957824829│   └── orientation_samples│       └── orientation_sample_1637143958005814├── 20211117-1140│   ├── depth│   │   └── depth_1637145649108790│   ├── filtered_cloud│   │   └── cloud_1637145649108790│   ├── joint_states│   │   └── joints_state_1637145648630977│   └── orientation_samples│       └── orientation_sample_1637145648831795└── 20211117-1205    ├── depth    │   └── depth_1637147164030135    ├── filtered_cloud    │   └── cloud_1637147164330388    ├── joint_states    │   └── joints_state_1637147163501574    └── orientation_samples        └── orientation_sample_1637147163655187

Cave reconstruction

- first_log_2cm_res_pointcloud-20231222.ply, contains the integrated pointcloud produced from the first of the logs.  

Coyote 3 Logs

The msgpack datasets can be imported using Python with the pocolog2msgpack library

The geometrical rover model of Coyote 3 is included in URDF format. It can be used in environment reconstruction algorithms which require the positions of the different sensors.

MP3

Includes exports of the log files used to compute the KPIs of the MP3. 

MP4

These logs were used to obtain the KPI values for the MP4. It is composed of the following archives:- log_coyote_02-03-2023_13-22_01-exp3.zip- log_coyote_02-03-2023_13-22_01-exp4.zip- log_coyote_02-09-2023_19-14_18_demo_skylight.zip- log_coyote_02-09-2023_19-14_20_demo_teleop.zip- coyote3_odometry_20230209-154158.0003_msgpacks.tar.gz- coyote3_odometry_20230203-125251.0819_msgpacks.tar.gz

Cave PLYs

Two integrated pointclouds and one trajectory produced from logs captured by Coyote 3 inside the cave:- Skylight_subsampled_mesh.ply- teleop_tunnel_pointcloud.ply- traj.ply

Example scripts to load the datasets

The repository https://github.com/Rauldg/corobx_dataset_scripts contains some example scripts which load some of the datasets.