CUDA-Accelerated Autonomous Navigation for Unitree Go2 Quadruped Robot

Features • Architecture • Quick Start • CUDA Acceleration • FAQ

This repository contains a complete autonomous navigation stack for the Unitree Go2 quadruped robot, featuring:

• MOLA LO - Modular Optimization Framework for LiDAR Odometry and mapping
• Terrain Analysis - Real-time traversability assessment
• Far Planner - GPU-accelerated visibility graph planning
• Local Planner - Reactive obstacle avoidance

LiDAR + IMU ──► MOLA LO ──► Terrain Analysis ──► Far Planner ──► Local Planner ──► Go2 Robot │ │ /lidar_odometry /terrain_map │ Terrain Analysis Ext │ /terrain_map_ext 

The terrain analysis nodes require tuning based on your robot's physical dimensions.

For a robot with base_link on the floor:

<!-- terrain_analysis.launch --> <param name="vehicleHeight" value="YOUR_ROBOT_HEIGHT" /> <param name="minRelZ" value="-0.5" /> <!-- Captures slight slopes --> <!-- terrain_analysis_ext.launch --> <param name="terrainUnderVehicle" value="-0.1" /> <!-- Small margin below floor -->

For a robot with base_link elevated (e.g., at center of mass):

<!-- terrain_analysis.launch --> <param name="vehicleHeight" value="YOUR_ROBOT_HEIGHT" /> <param name="minRelZ" value="-BASE_LINK_HEIGHT - 0.5" /> <!-- Goes below ground level --> <!-- terrain_analysis_ext.launch --> <param name="terrainUnderVehicle" value="-BASE_LINK_HEIGHT - 0.1" />

Slope is the nature of the slope.

The cyan lines show the visibility graph - navigation nodes that can "see" each other without obstacles. This computation is GPU-accelerated for real-time performance.

A modern, web-based interface is provided for high-level mission control and telemetry visualization.

• Interactive Map: Point-and-click interface to set navigation goals on the global map.
• Waypoint Management: specific locations can be saved, named, and recalled from a persistent list.
• Live Video Feed: Low-latency video streaming from the robot's camera.
• Teleoperation: Virtual joystick for manual control overrides.

The backend uses FastAPI to bridge ROS2 topics with the web frontend via WebSockets.

• Technology: Python (FastAPI, Uvicorn), AsyncIO.
• Real-time Data:
  • /ws/points → Stream of voxelized point clouds.
  • /ws/tf → Robot pose and map transforms.
  • /ws/video → JPEG-encoded camera stream.
• Persistence: waypoints.json stores user-defined locations.

To ensure smooth performance on the web interface, the high-density SLAM point cloud is downsampled before transmission.

• Node: voxel_grid_node (C++)
• Function: Applies a PCL VoxelGrid filter to reduce point count (~10x reduction) while preserving structural features.
• Topic Flow:
  • Input: /dlio/map_node/map (High density from SLAM)
  • Output: /registered_scan_o3d/voxelized (Optimized for Web UI)
• Configuration:
  • voxel_size: Default 0.1m (balances bandwidth vs. detail).

• ROS2 Humble
• CUDA 11.0+ (for GPU acceleration)
• Unitree Go2 SDK (for real robot)

# Clone the repository git clone https://github.com/Quadruped-dyn-insp/Go2_planner_suite.git cd Go2_planner_suite # Build all workspaces ./scripts/build.sh

MOLA_USE_FIXED_LIDAR_POSE=true \ MOLA_USE_FIXED_IMU_POSE=true \ MOLA_GENERATE_SIMPLEMAP=true \ MOLA_SIMPLEMAP_OUTPUT="myMap.simplemap" \ MOLA_SIMPLEMAP_MIN_XYZ=0.2 \ MOLA_LO_INITIAL_LOCALIZATION_METHOD="InitLocalization::PitchAndRollFromIMU" \ MOLA_DESKEW_METHOD="MotionCompensationMethod::IMU" \ MOLA_IMU_TOPIC="/livox/imu" \ MOLA_LIDAR_TOPIC="/livox/lidar" \ MOLA_TF_BASE_LINK="Head_upper" \ mola-lo-gui-rosbag2 /home/yasiru/Documents/rosbags/rosbag_004 sm2mm -i myMap.simplemap -o myMap.mm -p sm2mm_no_decim_imu_mls_keyframe_map.yaml mm-viewer myMap.mm -l libmola_metric_maps.so MOLA_GENERATE_SIMPLEMAP=true \ MOLA_SIMPLEMAP_OUTPUT="myMap.simplemap" \ MOLA_SIMPLEMAP_MIN_XYZ=0.2 \ MOLA_LO_INITIAL_LOCALIZATION_METHOD="InitLocalization::PitchAndRollFromIMU" \ MOLA_DESKEW_METHOD=MotionCompensationMethod::IMU \ MOLA_IMU_TOPIC="/livox/imu" \ MOLA_LIDAR_TOPIC="/livox/lidar" \ MOLA_TF_BASE_LINK="base_link" \ mola-lo-gui-rosbag2 \ /home/yasiru/Documents/Far_planner_test/rosbag2_2026_03_03-14_46_17

export MOLA_LO_PUBLISH_DESKEWED_SCANS=true source install/setup.bash ros2 launch mola_lidar_odometry ros2-lidar-odometry.launch.py \ start_active:=True \ publish_localization_following_rep105:=False \ start_mapping_enabled:=False \ lidar_topic_name:="/livox/lidar" \ imu_topic_name:="/livox/imu" \ mola_tf_base_link:="base_link" \ mola_deskew_method:="MotionCompensationMethod::IMU" \ mola_initial_map_mm_file:=$(pwd)/myMap.mm ros2 run tf2_ros static_transform_publisher 0 0 0 0 0 0 base_link livox_frame

ros2 bag play rosbag2_2026_03_03-15_06_01/ --loop source install/setup.bash && ros2 launch terrain_analysis terrain_analysis.launch source install/setup.bash && ros2 launch terrain_analysis_ext terrain_analysis_ext.launch source install/setup.bash && ros2 launch far_planner far_planner.launch.py

// Far Planner - Visibility checking __global__ void ComputeVisibilityConnections(...); __device__ bool IsEdgeCollidePolygons_GPU(...); __device__ bool doIntersect_GPU(...);

Go2_planner_suite/ ├── scripts/ │ ├── build.sh # Build all workspaces │ ├── setup.sh # Environment setup │ ├── launch.sh # Launch real robot │ └── sim.sh # Launch simulation ├── config/ # Global configuration ├── docs/ │ ├── images/ # Documentation images │ └── setup/ # Setup guides ├── tools/ # Utility scripts and debugging tools └── workspaces/ ├── autonomous_exploration/ # Mid-layer navigation framework │ ├── local_planner/ # Reactive obstacle avoidance │ ├── terrain_analysis/ # Local traversability mapping (near-field) │ ├── terrain_analysis_ext/ # Extended traversability mapping (far-field) │ └── go2_simulator/ # Gazebo simulation for Go2 ├── mola_lidar_odometry/ # MOLA LO framework ├── far_planner/ # CUDA-accelerated global planner │ ├── far_planner/ # Core visibility graph planner + CUDA │ └── boundary_handler/ # Obstacle boundary CUDA kernels └── pipeline_launcher/ # System orchestration & launch management 

sudo apt install ros-humble-pcl-ros ros-humble-tf2-ros \ ros-humble-nav-msgs ros-humble-geometry-msgs \ ros-humble-gazebo-ros-pkgs

conda deactivate # Use system Python for ROS # OR pip install <package> --target=/opt/ros/humble/lib/python3.10/site-packages

# Common remappings for real robot /velodyne_points: /livox/lidar /imu/data: /livox/imu /odom: /odom_dlio

*Terrain analysis uses CPU+OpenMP (CUDA version planned)

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

This project is licensed under the MIT License - see the LICENSE file for details.

• MOLA - Modular Optimization Framework for LiDAR Odometry
• FAR Planner - Planning algorithms
• Unitree Robotics - Go2 robot platform

_{Built for autonomous quadruped navigation}