Readme

English | 中文

Introduction

AgiBot X1 is a modular humanoid robot with high dof developed and open-sourced by AgiBot. It is built upon AgiBot's open-source framework AimRT as middleware and using reinforcement learning for locomotion control.

This project is about the accompanying software for AgiBot X1, including multiple functional modules such as model inference, platform driver, and software simulation.

For detailed tutorials on AimRT framework, please visit the AimRT official website.

Software Architecture Diagram

For detailed annotations of modules, please refer to the Development Guide.

Directory Structure

.
├── build.sh              # Build scripts
├── cmake                 # CMake scripts for build dependencies
│   ├── GetAimRT.cmake
│   ├── GetGTest.cmake
│   └── NamespaceTool.cmake
├── CMakeLists.txt        # Top-level CMakeLists.txt
├── format.sh             # Formatting scripts
├── README.md             # README document
├── doc                   # Development guide directory
├── src                   # Source code directory
│   ├── CMakeLists.txt    # CMakeLists.txt for the source code directory
│   ├── assistant         # ROS2 simulation and example project directory
│   ├── install           # Configuration script directory
│   ├── module            # Module directory
│   ├── pkg               # Deployment directory
│   └── protocols         # Protocol directory
└── test.sh               # Testing scripts

Running Instructions

Preparation for Launch

Install cmake (version 3.26 or above).
Install ONNX Runtime.

sudo apt update
sudo apt install -y build-essential cmake git libprotobuf-dev protobuf-compiler

git clone --recursive https://github.com/microsoft/onnxruntime

cd onnxruntime
./build.sh --config Release --build_shared_lib --parallel

cd build/Linux/Release/
sudo make install

Install ROS2 Humble and configure the environment variables. For detailed instructions, visit the ROS2 official website.
Install dependencies for the simulation environment.

sudo apt install jstest-gtk \
                 ros-humble-xacro \
                 ros-humble-gazebo-ros-pkgs \
                 ros-humble-gazebo-ros2-control \
                 ros-humble-joint-state-publisher \
                 ros-humble-joint-state-broadcaster \

Install the Linux realtime kernel patch if real-robot debugging is needed.

It will be very slow or even fail to download AimRT's numerous dependencies from default sources. Therefore, we provide environment variables DOWNLOAD_FLAGS for the download based on the Gitee source in url.bashrc. Simply source url.bashrc before running build.sh and add environment variable parameters for the Gitee source.

After completing the above steps, execute the following commands in the terminal:

source /opt/ros/humble/setup.bash
source url.bashrc

# Build
./build.sh $DOWNLOAD_FLAGS

# Test
./test.sh $DOWNLOAD_FLAGS

Launch Simulation

Prepare two terminals:

cd build/

# Terminal 1: Launch the inference module and joystick control module in simulation mode.
./run_sim.sh

# Terminal 2: Launch the Gazebo simulation program.
./run_gazebo.sh

Launch on Real Robot

Export your library path first, just need to execute once.

# Open "/etc/ld.so.conf" as root
sudo vi /etc/ld.so.conf

# Add this path to the end of "/etc/ld.so.conf"
/opt/ros/humble/lib
{YourProjectSource}/build/install/lib

# Refresh system env
sudo ldconfig

Ready to launch.

cd build/
./run.sh

Joystick Control

For specific control instructions, please refer to the Joystick Control Module.

License Agreement

The code provided in this project runs on the AimRT framework. This is research code subject to frequent changes, and cannot apply to all specific purposes. The source code is released under the MULAN license agreement.

Usage Instructions

If you have any questions or issues with this repository, please use the Issues.

Please do not contact us via email as we may not be able to respond. If you would like to contribute code, simply fork the repository (or if you are a collaborator, create a branch), make your changes, and submit a pull request to us.

DeepBehavier / agibot_x1_infer