Fourier-GRX

Welcome to the Fourier-GRX SDK, your gateway to controlling Fourier's humanoid robots! This guide will help you set up your environment, calibrate your humanoid robot, and run exciting demos. Let’s dive into the future of robotics! 🤖🚀

💻 System Requirement

• Operating System: Ubuntu 20.04 and up
• Python Version: Python 3.11

If you are first time using the robot, please set up the permission for IMU(HIPNUC IMU) and joysticks with permission_description

🚀 Environment Setup

Step 1: Install Conda

1. Download and install Miniconda: Follow the instructions on the Miniconda installation guide.

Step 2: Create and Activate Conda Environment

1. Create and activate the environment:

   [!NOTE] For now we only support Python 3.11.

   conda create -n grx-env python=3.11
   conda activate grx-env

   For more details, see the Conda user guide.

Step 3: Clone the Repository

1. Clone the official repo:

   git clone https://gitee.com/FourierIntelligence/wiki-grx-deploy.git

Step 4: Set Up Firewall

If you are using ufw, follow these steps to enable firewall access for the grx server to be able to automatically detect clients:

1. Enable firewall access:

   sudo ufw allow 7446/udp

2. Check the firewall status:

   sudo ufw status

Step 5: Install fourier-grx Library

1. Install the library:

   python -m pip install fourier-grx

2. Verify installation:

   grx --help

🛠️ Calibration

Caution

Ensure the sensor_offset.json file contains the correct absolute encoder values before running any code. Proper calibration is crucial for the robot’s operation and prevents potential damage. The grx run command reads the sensor_offset.json file from the directory where the server is started. Make sure this file is in the same directory when starting the server.

Calibrate Your Humanoid Robot's Absolute Encoder

1. Physical calibration could be down with the instructions shown in Physical_Calibration

2. Run the grx server:

   grx run ./config/config_GR1_T1.yaml --urdf-path ./urdf

3. Open a second terminal, activate the environment, and run:

   conda activate grx-env
   grx calibrate

Notice: Ensure that all calibration tools have been removed before proceeding with other operations.

This saves sensor offsets to sensor_offset.json.

🎛️ Usage

Start the grx Server

1. Run with the sample config file:

   grx run ./config/config_GR1_T1.yaml --urdf-path ./urdf

Run the Robot Client

1. Open a second terminal, activate the environment, and run:

   conda activate grx-env
   python demo_robot_client.py

Robot Client Options

• Enable: Enable the motor.
• Disable: Disable the motor.
• Set_Home: Calibrate all absolute encoders.
• Set_Gains: Set PD parameters.
• Reboot: Reboot all motors.
• Print_States: Print motor status.
• Move_to_Default: Move to default positions.
• Record: Save joint movements.
• Play: Replay recorded tasks.
• Abort: Stop current movements.
• List_Frames: List all URDF links.
• Get_Transform: Get transformation matrices.
• Exit: Exit the client panel.

🤖 Reinforcement Learning (RL) Demo

Warning

Ensure a safe position with ample space. Be ready to push emergency stop if needed.

Run RL Demos

1. Start the server:

   grx run ./config/config_GR1_TX.yaml --urdf-path ./urdf

2. Run the standing demo:

   python demo_nohla_stand.py --act

   Demo result:

   

3. Run the walking demo:

   First, you need to run demo_nohla_stand.py to get similar results as shown above. Then, place the robot on the ground and run the following code. Do not forget to ensure the robot is balanced before starting to walk.

   python demo_nohla_rl_walk.py --model-dir ./data/nohla_rl_walk --act

   Demo result:

   

Now your humanoid robot should be up and walking! 🦾🚶‍♂️

Enjoy exploring the fascinating world of humanoid robotics with fourier-grx! 🌍🤖✨