It is archived to preserve the code as it was at the time of the release of the paper referenced below. However, for a more up-to-date version, please see this repo.
This software is a portion of the paper Stronger Together: Air-Ground Robotic Collaboration Using Semantics, which can be found on arXiv and has been accepted to the IEEE Robotics and Automation Letters. Please cite this work if you find this code useful. The paper also has an accompanying video.
@ARTICLE{miller2022stronger,
author={Miller, Ian D. and Cladera, Fernando and Smith, Trey and Taylor, Camillo Jose and Kumar, Vijay},
journal={IEEE Robotics and Automation Letters},
title={Stronger Together: Air-Ground Robotic Collaboration Using Semantics},
year={2022},
volume={7},
number={4},
pages={9643-9650},
doi={10.1109/LRA.2022.3191165}
}
- OpenCV (Tested with 3.4)
- Eigen (Tested with 3.3)
- grid_map (Forked version) https://github.com/iandouglas96/grid_map
- GTSAM (Tested with 4.0.3)
- ROS1 (Tested with Melodic and Noetic)
You will also need a compiler with C++17 support (such as GCC 9 or greater).
Some data can be found here. To run:
roscore
rosparam set /use_sim_time true
roslaunch asoom asoom.launch
rosbag play --clock asoom_demo.bag
Then open rviz and load launch/viz.rviz.
If you would like to map semantics as well, you will need to also run ErfNet, as well as set the use_semantics parameter.
We use ORBSLAM3 for odometry with a wrapper found here.
However, odometry data is included in the provided bag file.
The settings used in our experiments are in asoom_iros_params.launch.