franka_gazebo_bringup

Important

Minimum necessary franka_description version is 0.3.0. You can clone franka_description package from https://github.com/frankarobotics/franka_description.

A project integrating Franka ROS 2 with the Gazebo simulator.

Launch RVIZ + Gazebo

Launch an example which spawns RVIZ and Gazebo showing the robot:

ros2 launch franka_gazebo_bringup visualize_franka_robot.launch.py

If you want to display another robot, you can define the robot_type:

ros2 launch franka_gazebo_bringup visualize_franka_robot.launch.py robot_type:=fp3

If you want to start the simulation including the franka_hand:

ros2 launch franka_gazebo_bringup visualize_franka_robot.launch.py load_gripper:=true franka_hand:='franka_hand'

Joint Velocity Control Example with Gazebo

Before starting, be sure to build franka_example_controllers and franka_description packages. franka_description must have the minimum version of 0.3.0.

colcon build --packages-select franka_example_controllers

Now you can launch the velocity example with Gazebo simulator.

ros2 launch franka_gazebo_bringup gazebo_franka_arm_example_controller.launch.py load_gripper:=true franka_hand:='franka_hand' controller:='joint_velocity_example_controller'

Keep in mind that the gripper joint has a bug with the joint velocity controller. If you are interested in controlling the gripper please use joint position interface.

Joint Position Control Example with Gazebo

To run the joint position control example you need to have the required software listed in the joint velocity control section.

Then you can run with the following command.

ros2 launch franka_gazebo_bringup gazebo_franka_arm_example_controller.launch.py load_gripper:=true franka_hand:='franka_hand' controller:='joint_position_example_controller'

Joint Impedance Control Example with Gazebo

Source your workspace.

source install/setup.sh

Then you can run the impedance control example.

ros2 launch franka_gazebo_bringup gazebo_franka_arm_example_controller.launch.py load_gripper:=true franka_hand:='franka_hand' controller:='joint_impedance_example_controller'

FR3 Duo Example with Gazebo

Before starting, be sure to build franka_example_controllers, franka_gazebo_bringup, gz_ros2_control and franka_description packages.

colcon build --packages-select franka_example_controllers franka_gazebo_bringup franka_description gz_ros2_control
source install/setup.bash

Now you can launch the FR3 duo example with Gazebo:

ros2 launch franka_gazebo_bringup gazebo_fr3_duo_example.launch.py

To launch with the complete sensor suite including the Vision and Manipulation Kit sensors, also build franka_vision_and_manipulation_kit:

colcon build --packages-select franka_vision_and_manipulation_kit
source install/setup.bash
ros2 launch franka_gazebo_bringup gazebo_fr3_duo_example.launch.py with_sensors:=true

Note

The sensor suite integrates:

  • franka_vision_and_manipulation_kit provides 3 sensors (2 wrist D405 cameras, 1 ZED Mini head camera)

All sensors are properly attached to the robot kinematic tree, ensuring proper simulation and sensor data streaming.

Important: When using with_sensors:=true, the Vision and Manipulation Kit includes Robotiq grippers.

Sensor Configuration with with_sensors:=true:

This command enables:

Vision and Manipulation Kit Sensors (from franka_vision_and_manipulation_kit):

  • 2x RealSense D405 cameras (left and right wrist cameras)

  • 1x ZED Mini camera (head camera)

Topics available:

Wrist cameras (Vision and Manipulation Kit):

  • /left_wrist_camera/image_raw, /right_wrist_camera/image_raw

Head camera (ZED Mini):

  • /head_camera/image_raw, /head_camera/image_raw/camera_info

Arguments:

  • with_sensors: If set to true, uses the complete sensor-enhanced description from the Vision and Manipulation Kit sensors (franka_vision_and_manipulation_kit) with Gazebo sensor plugins. Defaults to false.

  • world: SDF world filename inside franka_gazebo_bringup/worlds/ to load. Overrides the default world selection.

This will spawn two FR3 arms with gripper and wrist cameras, and start the joint impedance controller for both arms. RViz will also launch for visualization.

Mobile FR3 Duo Example with Gazebo

Before starting, be sure to build franka_example_controllers, franka_gazebo_bringup, gz_ros2_control and franka_description packages.

colcon build --packages-select franka_example_controllers franka_gazebo_bringup franka_description gz_ros2_control
source install/setup.bash

Now you can launch the mobile FR3 duo example with Gazebo:

ros2 launch franka_gazebo_bringup gazebo_mobile_fr3_duo_example.launch.py

Note

In simulation, a stub publisher sends false on /collision_detected with best effort QoS at 10 Hz, replacing the real self-collision node. This prevents the impedance controller from timing out on the collision topic.

To launch with the complete sensor suite including both the mobile platform sensors and the Vision and Manipulation Kit sensors, also build franka_mobile_sensors and franka_vision_and_manipulation_kit:

colcon build --packages-select franka_mobile_sensors franka_vision_and_manipulation_kit
source install/setup.bash
ros2 launch franka_gazebo_bringup gazebo_mobile_fr3_duo_example.launch.py with_sensors:=true

Note

The sensor suite integrates 10 sensors total from two packages:

  • franka_mobile_sensors provides 7 sensors (4 RGB cameras, 2 LiDARs, 1 IMU)

  • franka_vision_and_manipulation_kit provides 3 sensors (2 wrist D405 cameras, 1 ZED Mini head camera)

All sensors are properly attached to the robot kinematic tree, ensuring proper simulation and sensor data streaming.

Important: When using with_sensors:=true, the Vision and Manipulation Kit includes Robotiq grippers.

Sensor Configuration with with_sensors:=true:

This command enables BOTH sensor suites:

Mobile Platform Sensors (from franka_mobile_sensors):

  • 4x RealSense D455 cameras (front, rear, left, right)

  • 2x SICK nanoScan3 LiDARs (front, rear)

  • 1x OLV-IMU01 IMU

Vision and Manipulation Kit Sensors (from franka_vision_and_manipulation_kit):

  • 2x RealSense D405 cameras (left and right wrist cameras)

  • 1x ZED Mini camera (head camera)

Topics available:

Mobile platform cameras:

  • /camera_front/color/image_raw, /camera_rear/color/image_raw, etc.

Mobile platform LiDARs:

  • /lidar_front/scan, /lidar_rear/scan

Mobile platform IMU:

  • /imu/data

Wrist cameras (Vision and Manipulation Kit):

  • /left_wrist_camera/image_raw, /right_wrist_camera/image_raw

Head camera (ZED Mini):

  • /head_camera/image_raw, /head_camera/image_raw/camera_info

Arguments:

  • with_sensors: If set to true, uses the complete sensor-enhanced description with both mobile platform sensors (franka_mobile_sensors) and Vision and Manipulation Kit sensors (franka_vision_and_manipulation_kit) with Gazebo sensor plugins. Defaults to false.

  • world: SDF world filename inside franka_gazebo_bringup/worlds/ to load. Overrides the default world selection.

This will spawn the mobile base and two FR3 arms with gripper and wrist cameras, and start the joint impedance controller for both arms and cartesian velocity control for the mobile base. RViz will also launch for visualization. Select base_link to see the robot there.

Gravity Compensation in Simulation

Gravity is enabled globally in the Gazebo world, just like on the real robot. To keep the arms from collapsing under their own weight, franka_gazebo_bringup loads a gravity-compensation system plugin that computes the model-based gravity torque and applies it to the effort-controlled arm joints. This mirrors the real robot, where the master controller performs gravity compensation, so the zero-torque example controllers (for example the joint impedance controller) behave the same way in simulation as on hardware.

On the mobile platform (mobile_fr3_duo_v0_2), the vertical spine is a prismatic joint that would also drop under gravity. It is held at its initial height by the spine_joint_trajectory_controller, a JointTrajectoryController running on a position command interface, which is started automatically by the mobile example launch file.

You normally don’t need to configure any of this — it is wired up by the example launch files. Gravity being enabled is engine-independent and does not depend on a particular physics engine forwarding a gravity-disable flag.

Troubleshooting

If you experience that Gazebo can’t find your model files, try to include the workspace. E.g.

export GZ_SIM_RESOURCE_PATH=${GZ_SIM_RESOURCE_PATH}:/workspaces/src/