enav-planetary-dataset

Energy-aware navigation dataset for planetary rovers, collected at the Canadian Space Agency's Mars Emulation Terrain. This dataset amounts to more than 1,200 metres of driving and includes:

• Driving power consumption, solar irradiance on the rover's top plane and clear-sky direct irradiance estimates;
• Panoramic stereo color imagery and high-resolution forward-facing single-channel imagery;
• Tri-axial inertial measurements (acceleration & angular velocities);
• Individual wheel encoder data and corresponding 2D odometry estimates;
• Ground-truth GPS-based position and enhanced pose estimation from IMU, GPS and visual data fusion;
• Sun vector orientation relative to the rover's pose estimates;
• Geo-referenced maps of the environment (RGB mosaic, elevation/slope/aspect models).

Authors: Olivier Lamarre, Oliver Limoyo, Filip Marić, Dr. Jonathan Kelly
Affiliation: Space and Terrestrial Autonomous Robotic Systems (STARS) Laboratory, University of Toronto
Maintainer: Olivier Lamarre (email)

This dataset is described in details in our paper The Canadian Planetary Emulation Terrain Energy-Aware Rover Navigation Dataset, which was accepted for publication in The International Journal of Robotics Research. Please cite our work accordingly.

Overview

The dataset is further described in the project's official web page and can be downloaded from a dedicated IEEE DataPort page (requires creating a free account). The dataset is available in both rosbag and human-readable formats. This repository provides tools to interact with the rosbag files; we let users interested in the human-readable-formatted data develop their own utilities.

We provide a Docker container in which rosbags can be played and visualized using our custom ROS package, or parsed using our python utility scripts:

Lastly, this dataset also includes four different aerial maps of the test environment at a resolution of 0.2 meters per pixel: color, elevation, slope magnitude and slope orientation maps. Every map is georeferenced and is available in a .tif format. Tools to broadcast them as ROS topics (as grid_map messages) or simply load them using Python are also included:

Setup

As the dataset was originally collected with ROS Kinetic with a Ubuntu 16.04 machine, we provide a Docker container to play rosbags and/or extract specific data streams with Python.

1. Store downloaded rosbags in a single directory and the georeferenced maps in a dedicated subdirectory.

tree /path/to/dataset/dir

├── run1_base_new.bag
├── run2_base_new.bag
├── run3_base_new.bag
├── run4_base_new.bag
├── ...
├── maps
    ├── aspect_utm_20cm.tif
    ├── dem_utm_20cm.tif
    ├── mosaic_utm_20cm.tif
    └── slope_utm_20cm.tif

Note: obviously, you don't need to download all the rosbags - just the ones you need.

2. Clone the current repository and store the location of the downloaded dataset in a .env file in the project's root directory:

git clone https://github.com/utiasSTARS/enav-planetary-dataset.git
cd enav-planetary-dataset

echo "ENAV_DATASET_DIR='/path/to/enav_dataset/dir'" > .env

3. If not already done, install Docker. Then, set up X server permissions:

xhost +local:root

Run a container and enter it:

docker compose run --rm enav

Note that the dataset directory is now mounted at /enav_dataset in the container.

Rosbag interactions are done from inside the container:

• Playing rosbags and launching ROS visualization interfaces is documented in the enav_ros package's README file.

• Extracting data from rosbags with Python is documented in the enav_utilities package's README file.

Citation

@article{lamarre2020canadian,
   author = {Lamarre, Olivier and Limoyo, Oliver and Mari{\'c}, Filip and Kelly, Jonathan},
   title = {{The Canadian Planetary Emulation Terrain Energy-Aware Rover Navigation Dataset}},
   journal = {The International Journal of Robotics Research},
   year = {2020},
   doi = {10.1177/0278364920908922},
   URL = {https://doi.org/10.1177/0278364920908922},
   publisher={SAGE Publications Sage UK: London, England}
}