/mrca-mav

Collision avoidance for mavs in dynamic environments using model predictive control

Primary LanguageMATLABGNU General Public License v3.0GPL-3.0

Model Predictive Control for Multi-MAV Collision Avoidance in Dynamic Environments

This repository contains the code for the paper:

Chance-Constrained Collision Avoidance for MAVs in Dynamic Environments
Hai Zhu, and Javier Alonso-Mora
Published in [RA-L + ICRA 2019]. You can find the full-text paper here.

Please click in the image to see our video:

If you find this code useful in your research then please cite:

@article{Zhu2019RAL,
    title = {{Chance-Constrained Collision Avoidance for MAVs in Dynamic Environments}},
    author = {Zhu, Hai and Alonso-Mora, Javier},
    journal = {IEEE Robotics and Automation Letters},
    number = {2},
    volume = {4},
    pages = {776--783},
    publisher = {IEEE},
    year = {2019}
}

The authors would like to thank Embotech for providing a license of the FORCES PRO software.

Software Requirements

  • ROS installation
  • Ubuntu 16.04 (or 18.04)
  • MATLAB R2017b (or R2019b) with the Robotics System Toolbox
  • FORCES PRO software

Installation instructions

This set of instructions have been tested for Ubuntu 16.04 with ROS-Kinetic and MATLAB R2017b, and Ubuntu 18.04 with ROS-Melodic and MATLAB R2019b.

Running Simulations

  • Problem Set Up

    1. Launch a MATLAB instance and open initialize.m
    2. Setup the number of drones and dynamic obstacles
    3. Set cfg.modeSim as 1
    4. Set getNewSolver as 1 if a new mpc solver is required to be generated
  • Open Visualization

    1. Start a MATLAB instance
    2. Run the script run_visual.m
  • Open the Controller

    1. Start another MATLAB instance
    2. Run the script run_main_basic.m

Running Experiments

The code supports running experiments using the Parrot Bebop 2 quadrotors. Real-time state estimation of the quadrotors and moving obstacles are required. The following packages will be useful if you want to set up real-world experiments: