Skip to main content

Process Guide

We provide a complete workflow for implementing robot walking based on reinforcement learning, including the robot model, training strategy, validation strategy, and deployment strategy.

1. Prepare Robot Model​

URDF Model

  • Repository: Wiki-GRx-URDF
  • Description: Provides URDF format model files for the Fourier GRx series robots

MJCF Model

  • Repository: Wiki-GRx-MJCF
  • Description: Provides MJCF format model files for the Fourier GRx series robots

2. Train Walking Policy​

Isaac Gym Training Platform

  • Repository: Wiki-GRx-Gym
  • Description: Code implementation for training Fourier GRx robot walking policies in Isaac Gym

3. Validate Walking Policy​

MUJOCO Simulation Platform

  • Repository: Wiki-GRx-Mujoco
  • Description: Code implementation for validating Fourier GRx robot walking policies in Mujoco

4. Deploy to Real Robot​

  • Repositories: Wiki-GRx-Deploy, fourier_aurora_sdk
  • Description: Deploy the trained policy to the real Fourier GRx robot. Choose different SDKs based on different motion control frameworks: use Wiki-GRx-Deploy for the grx framework, and use fourier_aurora_sdk for the aurora framework.