The Company of iRobot
The TurtleBot4 platform consists of two main parts, the Create3 mobile base that controls the wheels, the interface buttons and docking/undocking. The other part is a Raspberry Pi 4 that runs ROS2 and oversees the LIDAR and the OAK-D camera. For simplicity, we have prepared the necessary system components upfront, so you will not need to access the onboard computers directly. When the TurtleBot turns on, the ROS2 topics required for control and reading data from sensors should automatically start, and you can access them via ROS2 from your computer if the network settings are correct.
The TurtleBot can only be booted up by placing it on the charging dock. Please follow the instructions below to ensure safe operation of the robot and avoid possible issues.
If the robot is powered off (lidar not spinning, LED ring is off):
- place the robot on the dock so both computers boot in sync
- wait for the beep (it may take a minute or two)
If the robot is already on the dock but the Raspberry Pi is not turned on (lidar not spinning, LED ring is on):
- remove the robot from the dock
- hold the Create3 power button for 10 seconds so the base powers down
- follow instructions for powering on
If the robot is running but not responsive (lidar is spinning, LED ring is on):
- remove the robot from the dock
- hold the Pi Shutdown button for a few seconds (the lidar should stop spinning)
- hold the Create3 power button for 10 seconds so the base powers down
- follow instructions for powering on
NOTE: The Create3 cannot be turned off while on the charging station, and the only way to turn it on when it's powered off is to place it on the dock. If the power is cut to the Pi 4 by turning off the Create3 before executing safe shutdown, it might corrupt the SD card. Always power off the Pi 4 first, before cutting power from the Create3.
If the buttons on the display do not respond or the display is off, it means that the Pi 4 is powered off, unless the lidar is spinning, in which case the ROS 2 nodes that handle the screen may not be running or the boot process hasn't finished yet.
OLED display: The first line shows the IP address. Buttons 3 and 4 select the action, button 1 confirms it, button 2 scrolls back to top. The top bar shows battery level and IP address
- Place the robot on the charging dock (green LED on the dock should turn on).
- If the robot was completely powered off before, wait until the chime sound.
- Hold the Pi Shutdown button for a few seconds (the lidar should stop spinning).
If the Raspberry Pi remains powered on, the robot will charge extremely slowly, so it's essential to turn it off for faster charging.
The Create3 base comes with a built-in LED ring around the main power button, which indicates the robot state. The other two buttons around it are unassigned by default and can be used for custom actions.
| State Description | Robot LED State |
|---|---|
| The Create3 is powered off, the Pi 4 power supply has been cut. In this state the robot can only be activated by placing it back on the dock. |  |
| The battery is charging, the robot is on the dock. |  |
| Normal operating state |  |
| Emergency stop (ESTOP) has been triggered, either due to hitting a bumper/IR/cliff switch or because there's no communication with the Pi 4. It can be reset by pressing the power button once, or through ROS |  |
| Battery is low, if you don't recharge the robot soon it will automatically power off. Place the robot back on the dock and power off the Pi 4 so it can recharge faster. |  |
The Turtlebot 4 consists of two DDS connected machines, the Pi 4 and the Create3 which are internally connected via USB-C. Both must be powered on for the robot to work properly.
Image source: Clearpath Robotics
As the robot turns on, it will connect to its dedicated color-coded router and network. Robots are split over multiple routers on different channels for more concurrent bandwidth.
If you are using your own laptop, first connect to the same network as the robot (e.g. for the robot Kili, connect to KiliWifi or KiliWifi_5GHz).
Lab workstations are already connected to one of the robot routers via Ethernet, and should be used with the corresponding robot.
The router to which the Gloin Turtlebot connects to, and the workstation that can be used with it
Next, you need to adjust your environment variables in the ~/.bashrc file. The ROS_DOMAIN_ID should match the one written on the robot, and RMW_IMPLEMENTATION needs to be set to rmw_cyclonedds_cpp.
The sticker with the ROS_DOMAIN_ID, different for each robot
The Cyclone middleware requires an xml config, specified as the CYCLONEDDS_URI environment variable. For more info about the cyclone config, see here, you might need to specify the network interface explicitly if using multiple ones e.g. <NetworkInterface name="wlan0" />. Lab workstations already have cyclone configured, if using your own laptop, you need to download and link the provided config.
Make sure your ~/.bashrc looks as follows:
export CYCLONEDDS_URI='/home/<your_user_name>/cyclonedds.xml'
export RMW_IMPLEMENTATION=rmw_cyclonedds_cpp
export ROS_DOMAIN_ID=<your_robot_domain_id>
export ROS_LOCALHOST_ONLY=0
Then, source ~/.bashrc, or re-open any terminals so the new .bashrc is properly sourced.
Optionally, restart the ros2 daemon just to be sure:
ros2 daemon stop; ros2 daemon start
This should be it!
To check if everything is working correctly, you can use the following checks:
To check if the robot is properly connected, you can inspect topics with:
ros2 topic list
You should see a lot of topics. If not, recheck:
- that the robot is charged and powered on
- your network settings
- try pinging the Raspberry Pi (
192.168.0.<your_robot_domain_id>) - your DDS config (
echo $RMW_IMPLEMENTATION) - your domain config (
echo $ROS_DOMAIN_ID)
See if you are receiving the /odom topic:
ros2 topic echo /odom
You should see a stream of data. If not, the Create3 base has not yet booted, wait for the chime sound. The /odom topic is necessary for localization and control of the robot.
See if you can move the robot with keyboard teleoperation:
ros2 run teleop_twist_keyboard teleop_twist_keyboard
You should be able to move the robot.
If you have passed all the checks, you can move on with building a map and navigating the robot.
See if the OAK-D camera is sending over the images by running:
ros2 topic echo --no-arr /oak/rgb/image_raw
ros2 topic hz /oak/rgb/image_raw
If you see data being published, you can then use the images in your scripts by changing the subscriber topic name.
The procedure for building a map is a little simpler than when using the simulation, we only need to start the SLAM and the keyboard teleoperation.
Please follow the official map making tutorial.
Once you have saved a map, we can have the robot navigate around the map.
Check the official navigation tutorial.
The transform system in ROS 2 is a standardized way to define and handle coordinate frames and transforms between them. It consists of two topics: /tf and /tf_static, as well as a collection of libraries for python and C++.
Every robot is defined by a collection of coordinate frames, with base_link being considered the parent frame for the current robot. Other common frames include:
base_footprint- typically below base link, where the robot meets the groundodom- the transformation between the location where the robot started and where it is now, based on wheel encoder datamap- the correction for odometry drift, usually calculated based on external landmarksworldorearth- localizes one or multiple robots on the world WGS84 system based on GNSS or other global data- sensor frames like
camera,imu,laser,sonar, etc. which reflect the position and rotation of sensors mounted on the real robot, so their data can be accurately transformed into other frames
When multiple robots are in the same TF graph, the conventional way to separate them is using namespacing, i.e. prepending a robot name to the frames. That way /robot1/base_link can be distinct from /robot2/base_link while using the same conventions or even be built from the same URDF file.
You can use the following command to collect and view currently available frames:
ros2 run tf2_tools view_frames
If you're running the TurtleBot simulator, the generated PDF should look something similar to the excerpt below, a full graph of all connected frames:
Check the official documentation page for more info.