DexRobot Python SDK

A comprehensive software development kit for robotic hand control, simulation, and manipulation. This project provides a complete stack for working with dexterous robotic hands, from low-level hardware control to high-level simulation and planning.

Getting Started

1. Make sure you have Git LFS installed:

sudo apt install git-lfs      # For Ubuntu, for example
git lfs install

2. Clone the repository:

git clone --recursive https://github.com/DexRobot/dexrobot_python_sdk.git

Make sure you have the --recursive flag turned on, so that all the submodules can be loaded. Alternatively, you can clone the submodules you need individually.

Project Components

1. pyzlg_dexhand

Low-level hardware interface and control library for the DexHand robotic system.

• Hardware communication via USB-CAN interface
• Protocol implementation for hand control
• ROS integration examples
• Visualization tools
• Comprehensive testing suite

2. dexrobot_kinematics

Kinematics library for robotic hand systems.

• Forward and inverse kinematics for hand configurations
• Support for both left and right hand variants
• Utilities for transformation and visualization
• Type-safe implementation with comprehensive testing

3. dexrobot_isaac

Isaac Sim integration and simulation environment.

• Complex scene creation and simulation
• Integration with OpenAI Gym-style environments
• Support for various manipulation tasks
• Reinforcement learning infrastructure
• Pre-built assets and environments:
  • Robotic hand models
  • Furniture and object models
  • Scene compositions
  • Training configurations

4. dexrobot_urdf

URDF (Unified Robot Description Format) models and utilities.

• Complete URDF descriptions for DexHand
• Mesh files for visualization and collision
• Utilities for URDF manipulation and analysis
• Both detailed and simplified hand models
• CAD exports in STEP format

5. dexrobot_mujoco

MuJoCo simulation environment and utilities.

• MuJoCo XML models for the DexHand
• Scene creation tools
• Real-time visualization
• Physics-based simulation
• Integration with ROS
• Rich set of demo environments and furniture models

6. ros_compat

ROS compatibility layer for platform-independent development.

• Abstract interface for ROS functionality
• Logging utilities
• Time handling
• Node implementation

Installation

Each component can be installed independently using pip:

# Install individual components
pip install ./pyzlg_dexhand
pip install ./dexrobot_kinematics
pip install ./dexrobot_isaac
pip install ./dexrobot_urdf
pip install ./dexrobot_mujoco
pip install ./ros_compat

Dependencies

• Python 3.6+
• ROS (optional, for ROS integration)
• MuJoCo (for physics simulation)
• Isaac Sim (for advanced simulation)
• USB-CAN adapter (for hardware control)

Documentation

Each component contains its own documentation in the docs directory. The documentation can be built using Sphinx:

cd <component_directory>
sphinx-build docs docs/_build

License

Each component may have its own license. Please refer to the LICENSE file in each component directory.

Project Structure

dexrobot_python_sdk/
├── pyzlg_dexhand/      # Hardware interface
├── dexrobot_kinematics/ # Kinematics library
├── dexrobot_isaac/     # Isaac Sim integration
├── dexrobot_urdf/      # Robot description files
├── dexrobot_mujoco/    # MuJoCo simulation
└── ros_compat/         # ROS compatibility layer

Contributing

Please refer to individual component READMEs for specific contribution guidelines.

Support

For issues and support:

• Hardware issues: Check pyzlg_dexhand documentation
• Simulation issues: Refer to dexrobot_isaac or dexrobot_mujoco
• Kinematics questions: See dexrobot_kinematics documentation