This repository include Python codes for the position control a UAV in a Gazebo simulation environment, using geometric controllers.
- Developed using Python
- Uses a geometric controller that works great with aggressive maneuvers
- Uses Gazebo as the physics engine
- Uses only minimal Gazebo/ROS codes so that if something breaks on ROS end in the future, it is easier to fix
- Has a nice GUI for controlling the UAV
- Can run your experiments, log data, and plot at the end of the trajectory at a click of a button
- Estimator, controller, and trajectory generators are in their own Python classes, if you need to test your own estimator, controller, or a trajectory, you only need to modify the respective class
- Python makes developing/debugging easier and shorter (no compiling)
- Can easily find modules or libraries for different tasks
- A geometric controller with decouppled-yaw attitude control is used
- The controller is published in:
@InProceedings{Gamagedara2019b, title={Geometric controls of a quadrotor uav with decoupled yaw control}, author={Gamagedara, Kanishke and Bisheban, Mahdis and Kaufman, Evan and Lee, Taeyoung}, booktitle={2019 American Control Conference (ACC)}, pages={3285--3290}, year={2019}, organization={IEEE} }
- Implementation of the same controller in C++ and Matlab can be found at https://github.com/fdcl-gwu/uav_geometric_control
- Make sure the desired trajectory generates the desired states required by your controller.
- Simply update the
Controller
class with your controller. - Make sure your modified controller class outputs the variable force-moment vector (
fM
).
- The estimator defined in the following paper is implemented here (except the sensor bias estimation terms):
@InProceedings{Gamagedara2019a, author = {Kanishke Gamagedara and Taeyoung Lee and Murray R. Snyder}, title = {Real-time Kinematics {GPS} Based Telemetry System for Airborne Measurements of Ship Air Wake}, booktitle = {{AIAA} Scitech 2019 Forum}, year = {2019}, month = {jan}, publisher = {American Institute of Aeronautics and Astronautics}, doi = {10.2514/6.2019-2377} }
- Matlab implementation of the above controller can be found at https://github.com/fdcl-gwu/dkf-comparison.
- Note that the Matlab implementation has a delayed Kalman filter that has not been implemented here. Only the non-delayed parts inside
DelayedKalmanFilter.m
is utilized here.
- Make sure your estimator provides the states for the UAV control, or if not create a separate estimator using the current estimator as the base class.
- Update the
Estimator
class with your estimator.
- Add your trajectory generation function in
Trajectory
class. - Replace that with an unused mode in the
calculate_desired
function inside theTrajectory
class.
- ROS: this repository has been developed using ROS Melodic, on Ubuntu 18.04.
- Python GTK libraries for GUI (not required if you opt to not to use the GUI)
sudo apt-get install python-gtk2-dev
- Python modules: these libraries must be installed in the system
- NumPy
- Pandas
- Matplotlib
- Clone the repositroy.
git clone https://github.com/fdcl-gwu/uav_simulator.git
- Update the submodules.
cd uav_simulator git submodule update --init --recursive
You only need to do the followings once (unless you change the Gazebo plugins)
- Make the pluging.
cd uav_simulator catkin_make
- Source the relevant directories (NOTE: you need to do this on every new terminal).
cd uav_simulator cd devel && source setup.bash && cd ../
- In the current terminal window, launch the Gazebo environment:
roslaunch uav_gazebo simple_world.launch
- Once the Gazebo is launched, run the rover code from a different rover terminal (if you already don't know, you may find tmux a life-saver):
If you change the Python code, simply re-run the Python code
python main.py
- Everytime you change the simulation environment, you have to kill the program,
catkin_make
and re-run it. - If you do not make any changes to the simulation environment, you only need to kill the Python program.
- The UAV will re-spawn at the position and orientation defined in
reset_uav()
inrover.py
when you run the Python code.