DJI Tello ROS2

• DJI Tello driver for ROS 2 based on DJITelloPy that uses the official SDK for the drone.
• Can be used to control multiple drones both using the swarm functionality (only for Tello EDU) or using multiple WLAN with regular Tello drones.
• This project was developed as a way of learning ROS 2 and evaluate the viability of moving other in progress projects from ROS 1 to ROS 2.
• Ii is recommended to update the Tello firmware to the latest version available
• Project workspace is divided into sub-workspaces that contain different logic.
  • tello package is the main package, includes access to the drone information, camera image and control.
  • tello_msg package defines custom messages to access specific Tello data.
    • Defines the TelloStatus, TelloID and TelloWifiConfig messages
  • tello_control package is a sample control package that displays the drone image and provides keyboard control.
    • T used for takeoff, L to land the drone, F to flip forward, E for emergency stop, WASD and arrows to control the drone movement.

• Bellow is the list of topics published and consumed by the tello package
• The list of published topics alongside their description and frequency. These topics are only published when some node subscribed to them, otherwise they are not processed.

• The list of topics subscribed by the node, these topics can be renamed in the launch file.

• The list of parameters used to configure the node. These should be defined on a launch file.

Camera Calibration

• To allow the drone to be used for 3D vision tasks, as for example monocular SLAM the camera should be first calibrated.
• A sample calibration file is provided with parameters captures from the drone used for testing but it is recommended to perform individual calibrations for each drone used.
• Calibration can be achieved using the camera_calibration package. Calibration pattern can be generated using the calib.io pattern generator tool.

ros2 run camera_calibration cameracalibrator --size 7x9 --square 0.16 image:=/image_raw camera:=/camera_info

• Take as many frame as possible and measure your check board grid size to ensure good accuracy in the process. When the process ends a calibrationdata.tar.gz will be created in the /tmp path.

Launch File

• Launch files in ROS2 are now defined using python code. To launch the main node of the project add the following code to your launch.py file.

Node(
    package='tello',
    executable='tello',
    namespace='/',
    name='tello',
    parameters=[
        {'tello_ip': '192.168.10.1'}
    ],
    remappings=[
        ('/image_raw', 'camera')
    ],
    respawn=True
)

Overheating Problems

• The motor drivers in the DJI Tello overheat after a while when the drone is not flying. To cool down the drivers i have removed the plastic section on top of the heat spreader (as seen in the picture).
• If you are comfortable with leaving the PCB exposed removing the plastic cover should result in even better thermals.
• If possible place the drone on top of an old computer fan or use a laptop cooler to prevent the drone from shutting down due to overheating.

Install

• To install the driver download the code from git, install dependencies using the script/install.sh script.
• After all dependencies are installed build the code and and install using colcon as usual or use the script/build.sh and script/run.sh to test the code.

Visual SLAM

• The drone is equipped with a IMU and a camera that can be used for visual SLAM in order to obtain the location of the drone and a map of the environment.
• ORB SLAM 2 is a monocular visual based algorithm for SLAM that can be easily integrated with the Tello drone using this package.
• The wrapper provided alongside with this repository is based on the alsora/ros2-ORB-SLAM2 project using the alsora/ORB_SLAM2 modified version of ORB Slam that does not depend on pangolin.
• To run the monocular SLAM node after installing all dependencies and building the package run.

ros2 run ros2_orbslam mono <VOCABULARY FILE> <CONFIG_FILE>

• The vocabulary file can be obtained from the ORB_SLAM2 repository ( ORB_SLAM2/Vocabulary/ORBvoc.txt).
• Sample configuration files can be found inside the package at orbslam2/src/monocular/config.yaml for monocular SLAM.

Setup ROS 2 Foxy

• Run the install script to setup the ROS 2 (Foxy Fitzroy) environment.
• Check the ROS2 Tutorials page to learn how to setup workspace and create packages.

Workspace

• To install dependencies of the packages available in a workspace directory src run rosdep install -i --from-path src --rosdistro foxy -y

• To build workspace you can use the command colcon build, some useful arguments for colcon build:

  • --packages-up-to builds the package you want, plus all its dependencies, but not the whole workspace (saves time)
  • --symlink-install saves you from having to rebuild every time you tweak python scripts
  • --event-handlers console_direct+ shows console output while building (can otherwise be found in the log directory)

Packages

• To create a new ROS2 package (C++ or Python) for development move to the src package and run

# CPP Package
ros2 pkg create --build-type ament_cmake --node-name <node_name> <package_name>

# Python Package
ros2 pkg create --build-type ament_python --node-name <node_name> <package_name>

Tools

• rqt_topic Used to monitor topics and their values in a list
• rqt_graph Draw the graph of connection between the currently active nodes and explore communication between them
• rviz Visualize topics in 3D space.

Bags

• Bags can be used to record data from topics that can be later replayed for off-line testing. Bags can be manipulated using the ros2 bag command. To

# Record a bag containing data from some topics into a file
ros2 bag record -o <bag_file_name> /turtle1/cmd_vel /turtle1/pose ...

# Check the content of a bag run the command
ros2 bag info <bag_file_name>

# Replay the content of some topics recorded into a bag file
 ros2 bag play <bag_file_name>

• To play ROS 1 bags in ROS 2 you will need to first install ROS 1, and the ROS bag adapter plugin. The the bags can be run using the command.

ros2 bag play -s rosbag_v2 <path_to_bagfile>

Camera calibration

• Calibration files provided were obtained using our test drone.
• To get your own calibration file use the ROS camera calibration tool

Ubuntu Based Linux Distros

• When installing on ubuntu based distros it might be required to change the distro codename so that the lsb_release -cs command returns the correct ubuntu base distribution.
• To change the output of the lsb_release command edit the /etc/os-release file. For ubuntu 20.04 the codename should be focal.
• Edit the file to contain the value UBUNTU_CODENAME=focal.

Windows Subsystem for Linux (WSL)

• Install WSL2 from the windows store or using the commands bellow, install Ubuntu 20.04 as the SO over the WSL overlay.

# Install WSL 2
Enable-WindowsOptionalFeature -Online -FeatureName Microsoft-Windows-Subsystem-Linux

# Enable WSL 2
dism.exe /online /enable-feature /featurename:VirtualMachinePlatform /all /norestart

# Check WSL version
wsl.exe --set-default-version 2
wsl -l -v

• Install a VcXsrv Windows X Server to be used as a X11 display server required to run GUI applications.
  • “Native opengl” unchecked
  • “Disable access control” checked
• Create a shortcut for VcXSrv with the following parameters

"C:\Program Files\VcXsrv\vcxsrv.exe" :0 -ac -terminate -lesspointer -multiwindow -clipboard -wgl -dpi auto

• To enable the access to the installed server add the display address to the .bashrc file

export DISPLAY="`grep nameserver /etc/resolv.conf | sed 's/nameserver //'`:0"
export LIBGL_ALWAYS_INDIRECT=0

• If you are using Visual Studio Code as and IDE you can configure for remote WSL development, allowing to debug code and interact with the WSL terminal.

• If you require CUDA acceleration you can also install NVidia CUDA drivers for WSL2

• If you get Clock skew detected. Your build may be incomplete. while compiling the code run the following commands or install the wsl-clock tool to automatically fix the clock drift problems.

sudo apt install ntpdate
sudo ntpdate time.windows.com