Name		Name	Last commit message	Last commit date
Environments		Environments
Media		Media
Models		Models
README.md		README.md
commands_file.txt		commands_file.txt
explore_params.yaml		explore_params.yaml
genesis_agent_test.py		genesis_agent_test.py
mapper_params.yaml		mapper_params.yaml
nav2.launch.py		nav2.launch.py
nav2_params.yaml		nav2_params.yaml
ppo_batch_train_code.py		ppo_batch_train_code.py
requirements.txt		requirements.txt
rsl_rl_model_test.py		rsl_rl_model_test.py
standRL.py		standRL.py
stand_ENV_batch.py		stand_ENV_batch.py
test.py		test.py
testing_push_up.py		testing_push_up.py
train_batch_Env_stand.py		train_batch_Env_stand.py
train_walk_env.py		train_walk_env.py
visualize_train_env.py		visualize_train_env.py
walk_env_test_controller.py		walk_env_test_controller.py
walk_train_with_rsl_lib.py		walk_train_with_rsl_lib.py

Repository files navigation

Genesis

This repository shows my Reinforcement Learning (RL) work for the Unitree Go2 robot.

I started this project about two months ago and now have results to share.

What I built:

Fast walking gait — about 5 m/s
Slow, stable gait — about 0.6 m/s

I also added domain randomization and terrain variations so the robot can handle:

Flat ground
Small bumps
Stairs and uneven surfaces

It stays balanced and walks well on these terrains.

I also added a camera on top of the robot's head.

About

A study of Quadruped Locomotion with Reinforcement Learning methoddsss.....

genesis ppo quadruped quadruped-robot quadruped-simulation reinforcement-learning robot-dog ros2

1 star

0 watching

Report repository

Releases

Packages

Contributors

Languages