unitree_go2_ros2

Unitree Go2 ROS2

Overview

This package provides a complete ROS 2 Jazzy integration for the Unitree Go2 quadrupedal robot using the CHAMP controller framework. It includes custom configuration packages and robot description models specifically adapted for ROS 2, enabling simulation, control, and autonomous operation capabilities.

About Unitree Go2

The Go2 is a quadrupedal robot manufactured by Unitree Robotics, designed for both research and commercial applications. It features powerful actuators, advanced sensor integration, and a robust mechanical design capable of navigating various terrains.

About CHAMP Controller

CHAMP (Coupled Hybrid Automata for Mobile Platforms) is an open-source development framework designed for quadrupedal robots. It provides a hierarchical control system that combines pattern modulation and impedance control techniques for efficient locomotion.

Features

• ✅ Complete ROS 2 Jazzy integration
• ✅ URDF model adapted to ROS 2 control framework
• ✅ Gazebo Harmonic simulation support
• ✅ Teleoperation using keyboard
• ✅ RVIZ visualization
• ✅ Integrated gait control and configuration
• ✅ Simulated sensors (in progress):
  • ✅ IMU
  • ✅ 2D LiDAR (Hokuyo)
  • ✅ 3D LiDAR (Velodyne)
  • ✅ 3D LiDAR (4D Lidar L1) (need some imporvments)
  • ✅ Mono Camera
  • ✅ Depth Camera (RealSense D455)
  • ✅ GPS (NavSat — simulates standard u-blox, ~0.5 m horizontal accuracy)
• ✅ Point cloud visualization in RVIZ
• ✅ Multiple sensor configurations available
• ❌ Full SLAM functionality (Coming soon)
• ❌ Navigation 2 integration (Coming soon)

System Requirements

• Ubuntu 24.04
• ROS 2 Jazzy
• Gazebo Sim Harmonic

Installation

1. Install ROS 2 Dependencies

sudo apt update
sudo apt install ros-jazzy-gazebo-ros2-control
sudo apt install ros-jazzy-xacro
sudo apt install ros-jazzy-robot-localization
sudo apt install ros-jazzy-ros2-controllers
sudo apt install ros-jazzy-ros2-control
sudo apt install ros-jazzy-velodyne
sudo apt install ros-jazzy-velodyne-description

2. Clone and Install CHAMP Controller and Go2 Simulation Packages

cd ~/ros2_ws/src
git clone https://github.com/khaledgabr77/unitree_go2_ros2

3. Install Dependencies

cd ~/ros2_ws
rosdep update
rosdep install --from-paths src --ignore-src -r -y

4. Build the Workspace

cd ~/ros2_ws
colcon build
source install/setup.bash

Usage

Gazebo Simulation

Launch the Gazebo simulation:

ros2 launch unitree_go2_sim unitree_go2_launch.py

Joint command interface

The joints are driven through ros2_control. By default the effort interface is used, matching joint_group_effort_controller in unitree_go2_sim/config/ros_control/ros_control.yaml. If needed, can be changed to position interface with:

ros2 launch unitree_go2_sim unitree_go2_launch.py command_interface:=position

Remember to switch the spawned controller in ros_control.yaml to a position controller as well, otherwise the controller manager will fail to claim the joints.

Leaving sensors out

All sensors are mounted by default. Each one can be left out of the robot description, which is useful when running headless or on a machine that struggles with the point cloud sensors:

Argument	Sensor
disable_camera	mono RGB camera
disable_d455	RealSense D455 RGBD camera
disable_lidar_l1	4D Lidar L1
disable_velodyne_lidar	Velodyne VLP-16

ros2 launch unitree_go2_sim unitree_go2_launch.py \
    disable_velodyne_lidar:=true disable_d455:=true

Watch Demo on YouTube

RVIZ Visualization

The package now includes both Velodyne 3D LiDAR and 4D Lidar L1 sensors. You can visualize the point cloud data in RVIZ:

Launch Gazebo with RVIZ:

ros2 launch unitree_go2_sim unitree_go2_launch.py rviz:=true

Velodyne Lidar and 4D Lidar L1

Velodyne Lidar Beams

4D Lidar L1 Beams

Velodyne Lidar and 4D Lidar L1 Beams

Mono Camera

Teleoperation

Control the robot using keyboard:

ros2 run teleop_twist_keyboard teleop_twist_keyboard

GPS Simulation

The robot includes a simulated GPS (NavSat) sensor mounted on top of the trunk, publishing at 10 Hz on /gps/fix as sensor_msgs/msg/NavSatFix.

The sensor models standard u-blox-level accuracy:

Axis	Noise (std dev)	Equivalent variance
Horizontal (lat/lon)	0.5 m	0.25 m²
Vertical (altitude)	0.8 m	0.64 m²

Verify it is working:

ros2 topic echo /gps/fix

Setting the geographic origin:

The world’s GPS origin is defined in unitree_go2_description/worlds/default.sdf under <spherical_coordinates>. Edit latitude_deg, longitude_deg, and elevation to match your actual test site before running navigation experiments:

<spherical_coordinates>
  <surface_model>EARTH_WGS84</surface_model>
  <world_frame_orientation>ENU</world_frame_orientation>
  <latitude_deg>YOUR_LAT</latitude_deg>
  <longitude_deg>YOUR_LON</longitude_deg>
  <elevation>YOUR_ALT_METERS</elevation>
  <heading_deg>0</heading_deg>
</spherical_coordinates>

Note: The position_covariance field in the bridged message will be all-zeros (COVARIANCE_TYPE_UNKNOWN) because the Gazebo→ROS bridge does not carry covariance. When integrating with robot_localization, set the covariance override in your EKF config or add a relay node that stamps diagonal covariance values matching the noise above.

Tuning Gait Parameters

The gait configuration for the robot is found in unitree_go2_sim/config/gait/gait.yaml. You can modify the following parameters:

Parameter	Description
Knee Orientation	How the knees should be bent (.>> .>< .<< .<>)
Max Linear Velocity X	Maximum forward/reverse speed (m/s)
Max Linear Velocity Y	Maximum sideways speed (m/s)
Max Angular Velocity Z	Maximum rotational speed (rad/s)
Stance Duration	How long each leg spends on the ground while walking
Leg Swing Height	Trajectory height during swing phase (m)
Leg Stance Height	Trajectory depth during stance phase (m)
Robot Walking Height	Distance from hip to ground while walking (m)
CoM X Translation	Offset to compensate for weight distribution
Odometry Scaler	Multiplier to calculated velocities for dead reckoning

Project Structure

• champ/: Core controllers and state estimation for CHAMP
• unitree_go2_description/: URDF models, meshes, and world files
• unitree_go2_sim/: Simulation launch files and configuration

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'feat: Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

Acknowledgements

This project builds upon and incorporates work from the following projects:

• Unitree Robotics - For the Go2 robot description (URDF model)
• CHAMP - For the quadruped controller framework
• CHAMP Robots - For robot configurations and setup examples
• unitree-go2-ros2 - ROS 2 package with gazebo classic

License