Guide to running the project

This is a simple tutorial walking through how to start a Gazebo simulation of Marvin, controlled via the keyboard. It assumes that marvin_gazebo, marvin_control, marvin_description are in your catkin workspace

It works on Windows via WSL 2 (Ubuntu) or on Pop!_OS and MacOS.

Repository structure

The repository is divided into the following parts:

• Docker: Dockerfile and automation scripts related to it are located in the root of the directory
• source: This directory is mounted as a catkin workspace to the Docker container, inside of it are the files required for running the robot.

In the source folder the structure is the following:

• marvin_description: Contains different models of the robot and launch files for showing them in rviz
• marvin_gazebo: Used for launching the gazebo simulation environment and storage of Gazebo worlds
• marvin_control: Different types of steering and their launch files. Simulation is launched in Gazebo
• marvin_plan: Functionalities related to planning

Each directory has its own README file that explains how it works.

Docker

If you are new to Docker, there is a simple tutorial available here or here. In the case of wanting to run GUI programs, follow this tutorial.

The Dockerfile contains everything that is required for running a ROS environment inside a container. We have provided convenience scripts for launching a container and building the images. The container my_ros is launched with 2 volumes attached to it: config, source. The config volume contains very bare-bones configuration files for using tmux inside the ROS container, another thing it contains is a bashrc file that source the catkin workspace and ros functionalities.

Furthermode, there are these convenience scripts:

• build.sh: builds the Docker container with image name my_ros
• run.sh: launches the container with required arguments, main important detail is support for display output. This allows for running graphical programs such as rviz, Gazebo, rqt. In the case of there being another container with name my_ros, it deletes the old one and launches a new one.
• connect.sh: in case there is a need to use a second terminal screen, you can use this script to join to the container. As mentioned above, tmux is installed in the container.

---

Work distribution

• Erik: Worked on the docker container and ensured that all required packages are present in Dockerfile. Helped with debugging issues around GUI support.
• Jonas: Automation scripts for the Dockerfile and for downloading Docker. Ensured that GUIs work on Windows.
• Milan: Used a virtual machine.
• Kirill: Used a virtual machine.

Steps for running a driving simulation

1. Install docker

2. Build the docker image by running

   ./build.sh

3. To create and run the my_ros container, run the script:

   ./run.sh

   Note: This will stop and recreate the container, if it already exists.

4. Initialize and build the project by running the following commands:

   cd ~/catkin_ws/
   sudo catkin build
   source ./devel/setup.bash

5. Run gazebo, rviz and topic publisher using

   roslaunch marvin_control marvin_skid_teleop.launch

   or

   roslaunch marvin_ackermann_gazebo_control marvin_ackermann_gazebo_control.launch

   This will start...

   • rviz which will show the positions of the robot in real time
   • gazebo which will be running the simulation
   • publishing to topic marvin/cmd_vel

Skid Steering

1. Now you need a second terminal instance for running teleop control, either via a TMUX session or by opening a new terminal and running:

   ./connect.sh

2. To initialize the second terminal, you don't need to rebuild, just run:

   cd ~/catkin_ws/
   source ./devel/setup.bash

3. To start the teleop control, run:

   rosrun teleop_twist_keyboard teleop_twist_keyboard.py cmd_vel:=/marvin/cmd_vel

   You will know this is successfull if you see small table with explanations for control of the vehicle.

or

1. For starting the movement planner, run:

   # Uses topics from skid steering
   rosrun marvin_plan hermit_curves_movement.py

Ackermann Steering

1. Open a new terminal, and navigate to the source folder, and run.

   catkin build
   source devel/setup.bash

2. Start publishing commands for movement.

   # Stop the robot.
   rostopic pub /cmd_vel geometry_msgs/Twist "{linear: {x: 0.0, y: 0.0}, angular: {z: 0.0}}" -1

   # Move straight
   rostopic pub /cmd_vel geometry_msgs/Twist "{linear: {x: 0.1}}"

   # Move straight while turning right
   rostopic pub /cmd_vel geometry_msgs/Twist "{linear: {x: 0.1}, angular: {z: 0.02}}"

   # Move straight while turning left
   rostopic pub /cmd_vel geometry_msgs/Twist "{linear: {x: 0.1}, angular: {z: -0.02}}"