LuSNAR Dataset

Introduction

This repository contains a lunar segmentation, navigation and reconstruction dataset LuSNAR based on multi-sensor (Stereo Camera, LiDAR, IMU) for autonomous exploration. The LuSNAR is a multi-task, multi-scene, and multi-label lunar dataset, it includes 9 lunar simulation scenes based on Unreal Engine and each scene is divided according to topographic relief and the density of objects.

our paper has been published on ArXiv. Paper

Scene Diagram

Data Diagram

The LuSNAR dataset includes:

• High-resolution stereo image pairs
• Panoramic semantic labels
• Dense depth maps
• LiDAR point clouds
• IMU data
• Rover pose data

Use Cases

The dataset can be used for comprehensive evaluation of autonomous perception and navigation systems:

• 2D/3D Semantic Segmentation
• Visual/LiDAR SLAM
• 3D Reconstruction

Availability

The LuSNAR dataset is available for download from CSTCloud. LuSNAR Dataset Download Password:fjZt

Dataset Structure

The LuSNAR dataset has a total size of 108GB, containing:

• 42GB of stereo image pairs
• 50GB of depth maps
• 356MB of semantic segmentation labels
• 14GB of single-frame point cloud data with semantic information

  ├── image1
  │   ├── RGB
  │   │   ├── timestamp1.png
  │   │   ├── timestamp2.png
  │   │   └── ...
  │   ├── Depth
  │   │   ├── timestamp1.png
  │   │   ├── timestamp2.png
  │   │   └── ...
  │   └── Label
  │       ├── timestamp1.png
  │       ├── timestamp2.png
  │       └── ...
  ├── image2
  │   ├── RGB
  │   │   ├── timestamp1.png
  │   │   ├── timestamp2.png
  │   │   └── ...
  │   ├── Depth
  │   │   ├── timestamp1.png
  │   │   ├── timestamp2.png
  │   │   └── ...
  │   └── Label
  │       ├── timestamp1.png
  │       ├── timestamp2.png
  │       └── ...
  ├── LiDAR
  │       ├── timestamp1.txt
  │       ├── timestamp2.txt
  │       └── ...
  ├── Rover_pose.txt
  └── IMU.txt

  The correspondence between the colors in semantic images and category numbers is as follows:

Category Number	Category	Color
0	Lunar regolith
1	Impact crater
2	Rock
3	Mountain
4	Sky

The correspondence between category numbers and their respective categories in the LiDAR point cloud data is as follows:

Diagram showing the orientation of the rover and different sensor coordinate systems(For details on the extrinsic parameters, please refer to the paper.): The extrinsic parameters of different sensors relative to the lunar rover body coordinate system are as follows:

• Parent: Body

  • Child: IMU

    • Translation: [1.000, 0.000, -1.500] // x, y, z
    • Rotation:
      • Quaternion: [0.000, 0.000, 0.000, 1.000] // qx, qy, qz, qw
      • RPY (radian): [0.000, -0.000, 0.000] // roll, pitch, yaw (rad)
      • RPY (degree): [0.000, -0.000, 0.000] // roll, pitch, yaw (degree)

  • Child: LiDAR

    • Translation: [1.000, 0.000, -1.500]
    • Rotation:
      • Quaternion: [0.000, 0.000, 0.000, 1.000]
      • RPY (radian): [0.000, -0.000, 0.000]
      • RPY (degree): [0.000, -0.000, 0.000]

  • Child: Left Camera

    • Translation: [1.000, -0.155, -1.500]
    • Rotation:
      • Quaternion: [0.406, 0.406, 0.579, 0.579]
      • RPY (radian): [1.223, -0.000, 1.571]
      • RPY (degree): [70.077, -0.000, 90.000]

  • Child: Right Camera

    • Translation: [1.000, 0.155, -1.500]
    • Rotation:
      • Quaternion: [0.406, 0.406, 0.579, 0.579]
      • RPY (radian): [1.223, -0.000, 1.571]
      • RPY (degree): [70.077, -0.000, 90.000]

File Format

LiDAR/timestamp.txt

| x [m] | y [m] | z [m] | category number |

Rover_pose.txt

| timestamp [ns] | p_RS_R_x [m] | p_RS_R_y [m] | p_RS_R_z [m] | q_RS_w [] | q_RS_x [] | q_RS_y [] | q_RS_z [] | v_RS_R_x [m s^-1] | v_RS_R_y [m s^-1] | v_RS_R_z [m s^-1] | b_w_RS_S_x [rad s^-1] | b_w_RS_S_y [rad s^-1] | b_w_RS_S_z [rad s^-1] | b_a_RS_S_x [m s^-2] | b_a_RS_S_y [m s^-2] | b_a_RS_S_z [m s^-2] |

IMU.txt

| timestamp [ns] | w_RS_S_x [rad s^-1] | w_RS_S_y [rad s^-1] | w_RS_S_z [rad s^-1] | a_RS_S_x [m s^-2] | a_RS_S_y [m s^-2] | a_RS_S_z [m s^-2] |