Reinforcement Learning for Drone Navigation and Pump Localization

This project is built as part of our master thesis, and aims to discover the theisability of controlling a drone using Reinforcement Learning. The main task of the drone is to map an entire unkown room, as well as finding points of interest (POI).

Authors

• @sandborg
• @maggedelle

Installation

This installation guide will explain how to run the PX4 part of the software, as well as the Turtlebot 3 demo.

Everything runs using ROS2 Humble on a Ubuntu 22.04 LTS

Needed for both demos

• ROS2 Humble which can be installed here.

• UPPAAL 5

• Stratego UTILS, installed by the following command

pip install strategoutil

PX4 Demo

• Install PX4 and Gazebo using this guide.

• Add SLAM Toolbox to your ROS2 workspace using this guide.

• Add ROS_GZ to your ROS2 workspace using this guide.

• Install PSUTIL running the command below

pip install psutil

• Install Pointcloud to LaserScan using this guide.

Start the DEMO by running the following code

cd ~/ws_ros2
source install/local_setup.bash

cd ~/Reinforcement-Learning-for-Drone-Navigation-and-Pump-Localization/stompc
python3 stompc.py


[OPTIONAL]
rviz2 -d ~/RL-Controlled-Drone-Using-UPPAAL-ROS/rviz2_cfg/basic_cfg.rviz

Turtlebot3 Demo

• You need a configured Turtlebot3 robot. Follow this guide to set it up (Remember to select Humble).

• Install SLAM Cartographer using their own guide. A guide to start Cartographer can be found here (Remember to select Humble).

• Checkout this branch.

• Run the following code

    cd ~/ws_ros2
    source install/local_setup.bash

    export TURTLEBOT3_MODEL=burger
    ros2 launch turtlebot3_cartographer cartographer.launch.py


    cd ~/Reinforcement-Learning-for-Drone-Navigation-and-Pump-Localization/stompc 
    python3 stompc_turtle.py

    [OPTIONAL]
    rviz2 -d ~/RL-Controlled-Drone-Using-UPPAAL-ROS/rviz2_cfg/basic_cfg.rviz

License

MIT