marr_gz

Ubuntu Noble Numbat (LTS 24.04) https://releases.ubuntu.com/noble/

ROS2 Jazzy (LTS 2024-2029) https://docs.ros.org/en/jazzy/

Gazebo Harmonic (LTS 2023-2028) https://gazebosim.org/docs/harmonic/

Install

• docker engine

https://docs.docker.com/engine/

Install docker engine (not docker Desktop!!!) (tested on v. 19.03, 20.10)

Usually, this should work on Ubuntu distributions

    sudo apt install docker.io

or install from binaries

    https://docs.docker.com/engine/install/binaries/

See also Post-installation steps for Linux. In particular, add your user to the docker group and log out and in again, before proceeding.

    sudo usermod -aG docker $USER

Install docker compose plugin (tested on v2.37)

https://docs.docker.com/compose/install/linux/

DOCKER_CONFIG=${DOCKER_CONFIG:-$HOME/.docker}
mkdir -p $DOCKER_CONFIG/cli-plugins
curl -SL https://github.com/docker/compose/releases/download/v2.37.0/docker-compose-linux-x86_64 -o $DOCKER_CONFIG/cli-plugins/docker-compose
chmod +x $DOCKER_CONFIG/cli-plugins/docker-compose

• For Nvidia drivers, install nvidia-docker2

Nvidia docker

Build

Standard procedure

Pull the system image and build the user image

cd docker
./build.bash

Building from scratch

In case of problems with building the image (due to updates in the apt keys, libraries, etc.), update the docker images and build from scratch (it will take time and bandwidth!)

cd docker
./build_from_scratch.bash

Run

Option 1: local graphic card

This option autodetects if nvidia drivers are present and uses nvidia runtime in docker

cd docker
./run.bash

In case of issues with running the GUI of the simulator, try sharing these files with the container

- /dev/dri/:/dev/dri/
- /tmp/.X11-unix:/tmp/.X11-unix
- $HOME/.Xauthority:/home/robot/.Xauthority

or disable X11 access control, running this command from the host OS

xhost +

To force using X11 and CPU (without nvidia drivers), use

cd docker
./run.bash cpu

Option 2: vnc

cd docker
./run.bash vnc

Use a browser on http://localhost:3000 to see the simulation.

Launch

Run from the container, i.e., from tmux windows opened by the run script.

Option 1. Using robot configuration

Copy marr_gz/config/launch_config_template.yaml in a new config file (e.g. my_launch.yaml)

Edit my_launch.yaml to configure your simulation.

Robot configuration parameters:

robot: 
  base: wheeled|reacher
  dof: 2..6            # dof of reacher base
  arms: false|true     # side arms for wheeled base
  pantilt: false|true  # pan-tilt for wheeled base (camera will be mounted here)
  imu: false|true      # for both wheeled and reacher
  lidar: false|true    # for both wheeled and reacher
  camera: false|rgb|rgbd  # for both wheeled and reacher
  arms_control_interface: effort|velocity|position   # reacher and wheeled arms

Example:

robot: 
  base: reacher
  dof: 3
  imu: false
  lidar: false
  camera: rgb
  arms_control_interface: position

Note: you can create many config launch files for different configurations.

Launch the simulator and ROS nodes

cd marr_gz/scripts
python marr_launch.py ../config/my_launch.yaml

Option 2. Using ros2 launch file

cd ros2_ws
colcon build
ros2 launch marr_gz marr_robot.launch.py <config_paraneters>

The list of configurable parameters is given by

ros2 launch marr_gz marr_robot.launch.py -s

Examples:

ros2 launch marr_gz marr_robot.launch.py robot_type:=wheeled

ros2 launch marr_gz marr_robot.launch.py robot_type:=wheeled arms:=True control_interface:=velocity

ros2 launch marr_gz marr_robot.launch.py robot_type:=reacher dof:=3 control_interface:=velocity

Models available

two-wheeled base

• wheeled: 2-wheeled robot

reacher arm

• reacher: 2 to 6 dof arm

actuators

• pan-tilt: for the wheeled base (cameras will be mounted here)
• amrs: two 1 dof arms for wheeled robot

sensors

• IMU
• lidar
• RGB camera
• RGBD camera

On the wheeled base, IMU and lidar are placed on the base, while cameras on the pan-tilt unit, if present, otherwise on a fixed support.

On the reacher base, sensors are placed on the end-effector.

Controllers

Available controllers:

• differential drive for wheeled robot
• effort|veocity|position controllers for arms

Check running controllers

ros2 control list_controllers

Test controllers

Launch a robot, as described above.

Test the controllers of the robot model. Run the test control (on another window terminal - use Ctrl-b c in tmux to create a new window)

cd marr_gz/script
python test_control.py

You should see all the joints moving.

Manual tests

wheeled robot

ros2 launch marr_gz marr_robot.launch.py robot_type:=wheeled

test

cd marr_gz/script
python test_control.py -fn ddrive 

wheeled robot with arms

position control interface

ros2 launch marr_gz marr_robot.launch.py robot_type:=wheeled arms:=True control_interface:=position

test

cd marr_gz/script
python test_control.py -fn arm_position 

reacher robot position control interface

ros2 launch marr_gz marr_robot.launch.py robot_type:=reacher dof:=2 control_interface:=position

test

cd marr_gz/script
python test_control.py -fn arm_position 

reacher robot velocity control interface

ros2 launch marr_gz marr_robot.launch.py robot_type:=reacher dof:=2 control_interface:=velocity

test

cd marr_gz/script
python test_control.py -fn arm_velocity

reacher robot effort control interface

ros2 launch marr_gz marr_robot.launch.py robot_type:=reacher dof:=2 control_interface:=effort

test

cd marr_gz/script
python test_control.py -fn arm_effort

Sensors

Use Rviz to see sensor data

ros2 run rviz2 rviz2

Predefined rviz files are available in the config folder, e.g.

cd marr_gz/config
ros2 run rviz2 rviz2 -d wheeled.rviz