Perception-Aware Image-Based Visual Servoing of Aggressive Quadrotor UAVs

Paper

If you use this code in an academic context, please cite the following

Qin, Chao, Qiuyu Yu, HS Helson Go, and Hugh H-T. Liu. "Perception-Aware Image-Based Visual Servoing of Aggressive Quadrotor UAVs." IEEE/ASME Transactions on Mechatronics (2023).

Bibtex:

@article{qin2023perception,
  title={Perception-Aware Image-Based Visual Servoing of Aggressive Quadrotor UAVs},
  author={Qin, Chao and Yu, Qiuyu and Go, HS Helson and Liu, Hugh H-T},
  journal={IEEE/ASME Transactions on Mechatronics},
  year={2023},
  publisher={IEEE}
}

I. Overview

This IBVS control algorithm is designed for aggressive quadrotors to ensure target visibility during agile flight.

Authors: Chao Qin, QIuyu Yu, Helso Go, and Hugh H.-T. Liu from FSC lab

Paper: Perception-Aware Image-Based Visual Servoing of Aggressive Quadrotor UAVs, Qin, Chao; Yu, Qiuyu; Go, Shing Hei Helson; Liu, Hugh H. -T., in IEEE/ASME Transactions on Mechatronics, 2023

Video Links: youtube

II. Installation

Get ROS

This framework is based on the Robot Operating System (ROS) and you therefore first need to install it (Desktop-Full Install) by following the steps described in the ROS Installation.

Get catkin tools

Get catkin tools with the following commands:

sudo apt-get install python-pip
sudo pip install catkin-tools

Create a catkin workspace

Create a catkin workspace with the following commands by replacing <ROS VERSION> with the actual version of ROS you installed:

cd
mkdir -p catkin_ws/src
cd catkin_ws
catkin config --init --mkdirs --extend /opt/ros/$ROS_DISTRO --merge-devel --cmake-args -DCMAKE_BUILD_TYPE=Release
cd src

Get catkin_simple

git clone https://github.com/catkin/catkin_simple

Get eigen_catkin

git clone https://github.com/ethz-asl/eigen_catkin

Get mav_comm

git clone https://github.com/ethz-asl/mav_comm

Get rotors_simulator

git clone https://github.com/ethz-asl/rotors_simulator

Clone this repository:

git clone https://github.com/FSC-Lab/fsc_aggressive_ibvs

Configure RotorS

mv fsc_aggressive_ibvs/utils/mav_fpv_sensor.gazebo rotors_simulator/rotors_description/urdf

We modify the weight of the simulated camera to zero for better visualization. At the <!-- VI-Sensor Macro --> insrc/rotors_simulator/rotors_description/urdf/component_snippets.xacro

change <mass value="0.13" /> to <mass value="0.0" />

Compile the source code

cd ..
catkin build

The source file to generetate MPC C codes using acados can be found in utils/pose_track_mpc.cpp

III. Target Tracking Simulation

source devel/setup.bash
roslaunch fsc_autopilot run_autopilot_rotors_planning.launch 

Click feature track to switch the control from position control to our IBVS

Specify the desired distance

And you will see the resulting flight trajectory in rviz. From the image in the left corner, we can see that our controller strives to maintain the target visual feature within the specified rectangular image bound.

IV. Acknowledgment

Our implementation is based on rpg_quadrotor_control and rpg_mpc.

We use ACADO as our NMPC solver.

We use RotorS for quadrotor simulation.