We propose an active localization correction system designed to balance the contradiction between environmental observation and mutual observation within the FoV-limited drone swarm, ensuring the swarm's flight safety.
[IROS 2024] Open source code for paper Preserving Relative Localization of FoV-Limited Drone Swarm via Active Mutual Observation. arXiv Preprint
Video Links: Bilibili.
We developed the code on Ubuntu 20.04 with ROS Noetic. Theoretically, it should also run on Ubuntu 18.04 and Ubuntu 16.04, but this has not been tested. Please install ROS(Robot Operating System) first following http://www.ros.org.
Run the following commands to setup:
git clone https://github.com/ZJU-FAST-Lab/Active-Relative-Localization.git
cd Active-Relative-Localization
catkin_make
We provide a launch file to run the simulation. You can run the simulation with the following command:
line formation
source devel/setup.bash && roslaunch ego_planner rviz.launch
source devel/setup.bash && roslaunch ego_planner line_formation.launch
x formation
source devel/setup.bash && roslaunch ego_planner rviz.launch
source devel/setup.bash && roslaunch ego_planner x_formation.launch
If you want to compare the difference between using and not using active mutual observation, you can change the parameter fsm/active_obs to true/false in src/plan_manage/launch/line_formation/advanced_param.xml
or src/plan_manage/launch/x_formation/advanced_param_x.xml
<param name="fsm/active_obs" value="true" type="bool"/>
The simulation and trajectory planning code are based on our previous work Ego-Planner-v2, and we use MINCO as our trajectory representation.