TurtleBot Lab Guide

• The booking system is available at https://cw.felk.cvut.cz/brute/student/reservation.php.
• Each student can book one robot for up to six 45min time slots (no more than 6 scheduled slots).
• Every two slots contain a 15min break, so that the slots are aligned with the school schedule hours.
• Note: All the robots are already reserved for ARO students during the ARO lab slots. There is no need to book the robots for your ARO lab slot, just take one.

DOs:

• Develop code on the desktops and make use of automatic synchronization of your home directory to the robots.
• Develop code on your laptop and use rsync to synchronize the code to the robot, but not every second.
• Develop code SSH'd on the robots in a text-mode editor like vim or nano (unless you're a Vim master, this is mostly good for hotfixes).
• You can also manage your code in a gitlab repository and synchronize code to the robot via the git repo (git push on your computer and git pull on the robot).
• Run visualization tools on the client machines (desktops, your laptop), not on the turtlebots via X11 tunneling.
• Run teleoperation code on the client machines (they wouldn't work when launched on the robot).
• Connect your laptop via Ethernet cable whenever it is possible.

DON'Ts:

• Don't copy large files over the wifi to or from the robot. You don't need any large files there. The Singularity image is already predownloaded in /local folder on the robots.
• Don't run silly find commands in the home directory of the robot. If you call it on a folder with many files, it literally kills the wifi for everybody.
• Don't run a graphical IDE via X11 tunelling (i.e. ssh -X … on the robot and run the IDE in there). The user experience from an IDE running over X11 tunnel is very bad. There might be exceptions, i.e. if you want to put a breakpoint in some code running on the robot, you might have to run the IDE directly on the robot in this way.
  • A better alternative is using the Remote interpreter option from Pycharm Professional (make sure to use the CTU free Professional license to enable remote development).
• Try to not connect your laptop via Wifi if possible.

The technical solution allowing to run a fully remotely managed fleet of robots and desktops on the very slow and unreliable network connection is actually quite interesting. If you're interested in details, read Turtlebot technical details (Czech only)

If you're wondering where you should run which part of the code, the rule is quite simple.

Drivers of the robot and sensor processing has to run on the robot (all the *_real.launch.xml files from homeworks).

Visualization and teleoperation tools have to run on your laptop or lab desktop PC (rviz, *_rviz.launch.xml, real_robot_*_teleop.launch.xml, …).

• If you do not need to work with the robots, but only need a computer that can run all the required software, you can also use the workstations in E132 - they are set up the same way.
• Login on a desktop computer (use your CTU username and CTU password).
• Set up an ARO ROS workspace according to Using ROS somewhere in your home directory.
• Select a TurtleBot from turtle01, turtle02, turtle08-turtle13, turn it on (after turning on the base, wait 10 seconds to check whether the PC power-on light went on; if not, press the power-on button to start the computer).
• Wait until the robot boots into a graphical login screen (some robots' screens are broken and stay black all the time, e.g. turtle01).
• Both the desktop and robot use the same network home directory. The workspace you configured above is directly available on the robot in the same path.
• In a terminal window, SSH to your robot (use hostname of your robot, your CTU username and password) :

  # on the desktop PC
  ssh -YC your_ctu_username@turtle08 # type in your CTU password

• On the robot, if you run the code for the first time on the robot, clone the ARO deploy repository

  # on robot:
  git clone https://gitlab.fel.cvut.cz/robolab/deploy.git

• On the robot, run robot drivers (and your localization/control code) from within your workspace:

  # on robot:
  ~/deploy/scripts/start_apptainer_aro tmux
  # (the console prompt should start with ARO)
  ros2 run rmw_zenoh_cpp rmw_zenohd  # keep zenohd running and open a new tmux tab with ''Ctrl-b+c'' or ''Ctrl-a+c''
  ros2 launch aro_exploration real_robot.launch.xml  # in a new tmux tab, start the base robot driver

• Still on the robot, open a new tmux window and run the code you want to test:

  # on robot:
  ros2 launch aro_exploration aro_localization_real.launch.xml run_driver:=false  # in a new tmux tab, start the homework-specific launch file

• Visualize, process robot data, develop or teleoperate the robot from the desktop computer. Run on a desktop terminal:

  # on the desktop PC
  # Replace turtleXY with the actual hostname of your turtlebot or with its IPv4 address
  export ZENOH_CONFIG_OVERRIDE='mode="client";connect/endpoints=["tcp/turtleXY:7447"]'
  ~/deploy/scripts/start_apptainer_aro
  # (the console prompt should start with ARO)
  ros2 topic list  # List all topics
  ros2 run rqt_tf_tree rqt_tf_tree  # Show TF frames
  ros2 run rqt_graph rqt_graph  # Show all nodes and topics
  
  rviz2  # 3D visualizer
  ros2 launch aro_exploration aro_localization_rviz.launch.xml  # 3D visualizer preconfigured for a specific homework
  
  ros2 launch aro_exploration real_robot_key_teleop.launch.xml  # Control the robot via keyboard
  ros2 launch aro_exploration real_robot_joy_teleop.launch.xml  # Control the robot via the connected gamepad
  
  code .  # Start VSCode inside the Singularity container
  pycharm  # Start Pycharm inside the Singularity container
  subl  # Start Sublime editor inside the Singularity container

• Try controlling the robot with your aro_reactive_control package, e.g. ros2 launch aro_exploration aro_slam_real.launch.xml run_driver:=false reactive_control:=true # run on the robot!.
• Try your exploration pipeline once it's ready.

• Set up an ARO ROS workspace according to Using ROS somewhere on your laptop.
• Select a TurtleBot from turtle01, turtle02, turtle08-turtle13, turn it on (after turning on the base, wait 10 seconds to check whether the PC power-on light went on; if not, press the power-on button to start the computer).
• Wait until the robot boots into a graphical login screen (some robots' screens are broken and stay black all the time, e.g. turtle01).
• Connect your laptop to an Ethernet cable (preferred), or to the lab wifi: SSID: turtlebots, password: TurtlesDC .
  • If you connect via cable, turn off the Wifi.
  • If you set up the wifi, configure it to not connect automatically. It will save you some trouble when you'll want to connect via Ethernet later.
• On your computer, test you can reach the robot: ping turtle08
  • Do not run this command inside Apptainer, it won't work (instead, run it in a console of your own operating system)
  • If you get name resolution error from ping and you are connected via cable, make sure your wifi is disconnected. If you're connected only via Wifi, check that you are connected to the turtlebots wifi. If you still cannot ping the robot by its hostname, try pinging its IP address, e.g. 192.168.65.21 (it is written on the robot).
    • If pinging the IP doesn't work, your connection is wrong.
    • If pinging the IP works and pinging the hostname does not, then it is a DNS resolution problem. You can ignore this problem and use the robot's IP address everywhere instead of its hostname.
• In a terminal window, SSH to your robot (use hostname of your robot and your CTU username and password) :

  # on your laptop
  ssh -XYC your_ctu_username@turtle08

• Connected via SSH to the robot, set up an ARO ROS workspace according to Using ROS somewhere in your home directory.
• Use rsync to synchronize files from your computer to the robot:

  # on your laptop
  rsync -aAP /your/path/to/workspace/src/ your_ctu_username@turtle08:workspace/src
  # The ending / after the first path is important, as well as not having a / after the 2nd path

  • Alternatively, if your IDE supports automatic deployment over SSH/SCP/RSync, you can set the synchronization in the IDE.
• On the robot, run robot drivers (and your localization/control code) from within your workspace:

  # on robot
  ~/deploy/scripts/start_apptainer_aro tmux
  # (the console prompt should start with ARO)
  ros2 run rmw_zenoh_cpp rmw_zenohd  # keep zenohd running and open a new tmux tab with ''Ctrl-b+c''
  ros2 launch aro_exploration real_robot.launch.xml  # in a new tmux tab, start the base robot driver
  ros2 launch aro_exploration aro_localization_real.launch.xml run_driver:=false  # in a new tmux tab, start the homework-specific launch file

• Then you can use the same visualization/teleoperation commands as described above

To configure passwordless login to the robots, run the following commands on the desktop or on your laptop:

ssh-keygen  # Only if you don't yet have an SSH key yet (stored in ~/.ssh/id_rsa)
ssh-copy-id your_ctu_username@turtle08

• General troubleshooting for working with Apptainer: Common problems
• Most things seem to work but I cannot teleoperate the robot using gamepad or keyboard.
  • Make sure you launch the teleoperation launch file on the desktop/laptop and not on the robot.
  • If you are on a lab desktop computer, the previous should fix it. If not, contact the lab maintainer.
  • It is possible the robot cannot connect to your laptop. SSH to the robot and try ping my-computer-hostname (outside of Apptainer). You can figure out the hostname of your computer e.g. by calling command hostname on your laptop. If the ping command does not succeed, disable wifi on your computer, connect cable, wait a few minutes and reboot your computer. This should resolve the issue.
• My computer doesn't see any topics from the robot (e.g. ros2 topic list is almost empty or cannot communicate with master):
  • Make sure you have executed export ZENOH_CONFIG_OVERRIDE='mode="client";connect/endpoints=["tcp/turtleXY:7447"]' (with the correct turtle name) in the console where you try to communicate with the robot.
  • Call ros2 daemon stop on your computer.
  • Make sure you are connected either via an Ethernet wire in the lab network or you are connected to the turtlebots wifi.
  • Make sure you can ping turtle08 (do that from outside the Apptainer in a normal system terminal).
    • If you cannot do this, check your /etc/resolv.conf file. Run ls -la /etc/resolv.conf. The file should be a symbolic link. If it's not, you have configured your system to use a custom DNS server and you have to revert this configuration. If the file is a symlink, try cat /etc/resolv.conf. If it contains entries like 127.0.1.1, 127.0.0.1 or 192.168.65.5, that's okay. If it contains other entries, check your wired or wireless connection settings and configure it for automatic DNS.
• ROS doesn't work in tmux (and I use a custom ~/.tmux.conf)
  • Add the following to your .tmux.conf: set-option -g default-shell /bin/bash
• Sometimes, it happens that the Apptainer image on the robots is truncated to 0 bytes or damaged. When this happens, start_apptainer_aro fails with various error messages like “image format not recognized” and similar.
  • To resolve this issue, just reboot the robot.
• If your password does not work for the SSH, check if you have a VPN running and disable it if you do.