robot_lab

Overview

robot_lab is a RL extension library for robots, based on IsaacLab. It allows you to develop in an isolated environment, outside of the core Isaac Lab repository.

Important

This repository currently depends on IsaacLab v2.0.0 or higher. If you prefer to use IsaacLab v1.4.1, please use it with v1.1 of this repository.

Note

If you want to run policy in gazebo or real robot, please use rl_sar project.

Discuss in Github Discussion or Discord.

Installation

Install Isaac Lab by following the installation guide. We recommend using the conda installation as it simplifies calling Python scripts from the terminal.
Clone this repository separately from the Isaac Lab installation (i.e. outside the IsaacLab directory):
```
git clone https://github.com/fan-ziqi/robot_lab.git
```
Using a python interpreter that has Isaac Lab installed, install the library
```
python -m pip install -e source/robot_lab
```
Verify that the extension is correctly installed by running the following command to print all the available environments in the extension:
```
python scripts/tools/list_envs.py
```

Set up IDE (Optional, click to expand)

To setup the IDE, please follow these instructions:

Run VSCode Tasks, by pressing Ctrl+Shift+P, selecting Tasks: Run Task and running the setup_python_env in the drop down menu. When running this task, you will be prompted to add the absolute path to your Isaac Sim installation.

If everything executes correctly, it should create a file .python.env in the .vscode directory. The file contains the python paths to all the extensions provided by Isaac Sim and Omniverse. This helps in indexing all the python modules for intelligent suggestions while writing code.

Setup as Omniverse Extension (Optional, click to expand)

We provide an example UI extension that will load upon enabling your extension defined in source/robot_lab/robot_lab/ui_extension_example.py.

To enable your extension, follow these steps:

Add the search path of your repository to the extension manager:
- Navigate to the extension manager using Window -> Extensions.
- Click on the Hamburger Icon (☰), then go to Settings.
- In the Extension Search Paths, enter the absolute path to robot_lab/source
- If not already present, in the Extension Search Paths, enter the path that leads to Isaac Lab's extension directory directory (IsaacLab/source)
- Click on the Hamburger Icon (☰), then click Refresh.
Search and enable your extension:
- Find your extension under the Third Party category.
- Toggle it to enable your extension.

Docker setup

Click to expand

Building Isaac Lab Base Image

Currently, we don't have the Docker for Isaac Lab publicly available. Hence, you'd need to build the docker image for Isaac Lab locally by following the steps here.

Once you have built the base Isaac Lab image, you can check it exists by doing:

docker images

# Output should look something like:
#
# REPOSITORY                       TAG       IMAGE ID       CREATED          SIZE
# isaac-lab-base                   latest    28be62af627e   32 minutes ago   18.9GB

Building robot_lab Image

Following above, you can build the docker container for this project. It is called robot-lab. However, you can modify this name inside the docker/docker-compose.yaml.

cd docker
docker compose --env-file .env.base --file docker-compose.yaml build robot-lab

You can verify the image is built successfully using the same command as earlier:

docker images

# Output should look something like:
#
# REPOSITORY                       TAG       IMAGE ID       CREATED             SIZE
# robot-lab                        latest    00b00b647e1b   2 minutes ago       18.9GB
# isaac-lab-base                   latest    892938acb55c   About an hour ago   18.9GB

Running the container

After building, the usual next step is to start the containers associated with your services. You can do this with:

docker compose --env-file .env.base --file docker-compose.yaml up

This will start the services defined in your docker-compose.yaml file, including robot-lab.

If you want to run it in detached mode (in the background), use:

docker compose --env-file .env.base --file docker-compose.yaml up -d

Interacting with a running container

If you want to run commands inside the running container, you can use the exec command:

docker exec --interactive --tty -e DISPLAY=${DISPLAY} robot-lab /bin/bash

Shutting down the container

When you are done or want to stop the running containers, you can bring down the services:

docker compose --env-file .env.base --file docker-compose.yaml down

This stops and removes the containers, but keeps the images.

Try examples

Note

If you want to control a SINGLE ROBOT with the keyboard during playback, add --keyboard at the end of the play script.

Key bindings:
====================== ========================= ========================
Command                Key (+ve axis)            Key (-ve axis)
====================== ========================= ========================
Move along x-axis      Numpad 8 / Arrow Up       Numpad 2 / Arrow Down
Move along y-axis      Numpad 4 / Arrow Right    Numpad 6 / Arrow Left
Rotate along z-axis    Numpad 7 / Z              Numpad 9 / X
====================== ========================= ========================

Quadruped

Unitree A1

# Train
python scripts/rsl_rl/base/train.py --task RobotLab-Isaac-Velocity-Rough-Unitree-A1-v0 --headless
# Play
python scripts/rsl_rl/base/play.py --task RobotLab-Isaac-Velocity-Rough-Unitree-A1-v0

Unitree Go2

# Train
python scripts/rsl_rl/base/train.py --task RobotLab-Isaac-Velocity-Rough-Unitree-Go2-v0 --headless
# Play
python scripts/rsl_rl/base/play.py --task RobotLab-Isaac-Velocity-Rough-Unitree-Go2-v0

Unitree B2

# Train
python scripts/rsl_rl/base/train.py --task RobotLab-Isaac-Velocity-Rough-Unitree-B2-v0 --headless
# Play
python scripts/rsl_rl/base/play.py --task RobotLab-Isaac-Velocity-Rough-Unitree-B2-v0

Anymal D

# Train
python scripts/rsl_rl/base/train.py --task RobotLab-Isaac-Velocity-Rough-Anymal-D-v0 --headless
# Play
python scripts/rsl_rl/base/play.py --task RobotLab-Isaac-Velocity-Rough-Anymal-D-v0

Wheeled

Unitree Go2W

# Train
python scripts/rsl_rl/base/train.py --task RobotLab-Isaac-Velocity-Rough-Unitree-Go2W-v0 --headless
# Play
python scripts/rsl_rl/base/play.py --task RobotLab-Isaac-Velocity-Rough-Unitree-Go2W-v0

Unitree B2W

# Train
python scripts/rsl_rl/base/train.py --task RobotLab-Isaac-Velocity-Rough-Unitree-B2W-v0 --headless
# Play
python scripts/rsl_rl/base/play.py --task RobotLab-Isaac-Velocity-Rough-Unitree-B2W-v0

Humanoid

FFTAI GR1T1

# Train
python scripts/rsl_rl/base/train.py --task RobotLab-Isaac-Velocity-Rough-FFTAI-GR1T1-v0 --headless
# Play
python scripts/rsl_rl/base/play.py --task RobotLab-Isaac-Velocity-Rough-FFTAI-GR1T1-v0

FFTAI GR1T2

# Train
python scripts/rsl_rl/base/train.py --task RobotLab-Isaac-Velocity-Rough-FFTAI-GR1T2-v0 --headless
# Play
python scripts/rsl_rl/base/play.py --task RobotLab-Isaac-Velocity-Rough-FFTAI-GR1T2-v0

Unitree H1

# Train
python scripts/rsl_rl/base/train.py --task RobotLab-Isaac-Velocity-Rough-Unitree-H1-v0 --headless
# Play
python scripts/rsl_rl/base/play.py --task RobotLab-Isaac-Velocity-Rough-Unitree-H1-v0

Unitree G1

# Train
python scripts/rsl_rl/base/train.py --task RobotLab-Isaac-Velocity-Rough-Unitree-G1-v0 --headless
# Play
python scripts/rsl_rl/base/play.py --task RobotLab-Isaac-Velocity-Rough-Unitree-G1-v0

AMP for Quadruped

The code for AMP training refers to AMP_for_hardware

Unitree A1

# Retarget motion files
python source/robot_lab/robot_lab/third_party/amp_utils/scripts/retarget_kp_motions.py
# Replay AMP data
python scripts/rsl_rl/amp/replay_amp_data.py --task RobotLab-Isaac-Velocity-Flat-Amp-Unitree-A1-v0
# Train
python scripts/rsl_rl/amp/train.py --task RobotLab-Isaac-Velocity-Flat-Amp-Unitree-A1-v0 --headless
# Play
python scripts/rsl_rl/amp/play.py --task RobotLab-Isaac-Velocity-Flat-Amp-Unitree-A1-v0

HandStand

Supports four directions of handstand, use handstand_type in the configuration file to switch.

Unitree A1

# Train
python scripts/rsl_rl/base/train.py --task RobotLab-Isaac-Velocity-Flat-HandStand-Unitree-A1-v0 --headless
# Play
python scripts/rsl_rl/base/play.py --task RobotLab-Isaac-Velocity-Flat-HandStand-Unitree-A1-v0

Note

You can change Rough to Flat in the above configs.
Record video of a trained agent (requires installing ffmpeg), add --video --video_length 200
Play/Train with 32 environments, add --num_envs 32
Play on specific folder or checkpoint, add --load_run run_folder_name --checkpoint model.pt
Resume training from folder or checkpoint, add --resume --load_run run_folder_name --checkpoint model.pt

Add your own robot

For example, to generate Unitree A1 usd file:

python scripts/tools/convert_urdf.py source/robot_lab/data/Robots/Unitree/A1/a1_description/urdf/a1.urdf source/robot_lab/data/Robots/Unitree/A1/a1.usd --merge-join

Note

There is a problem with the urdf conversion of the current Isaacsim version. Please checkout your IsaacLab to v1.4.1 and use scripts/tools/convert_urdf_v1.py for conversion.

Check import_new_asset for detail

Using the core framework developed as part of Isaac Lab, we provide various learning environments for robotics research. These environments follow the gym.Env API from OpenAI Gym version 0.21.0. The environments are registered using the Gym registry.

Each environment's name is composed of Isaac-<Task>-<Robot>-v<X>, where <Task> indicates the skill to learn in the environment, <Robot> indicates the embodiment of the acting agent, and <X> represents the version of the environment (which can be used to suggest different observation or action spaces).

The environments are configured using either Python classes (wrapped using configclass decorator) or through YAML files. The template structure of the environment is always put at the same level as the environment file itself. However, its various instances are included in directories within the environment directory itself. This looks like as follows:

source/robot_lab/tasks/locomotion/
├── __init__.py
└── velocity
    ├── config
    │   └── unitree_a1
    │       ├── agent  # <- this is where we store the learning agent configurations
    │       ├── __init__.py  # <- this is where we register the environment and configurations to gym registry
    │       ├── flat_env_cfg.py
    │       └── rough_env_cfg.py
    ├── __init__.py
    └── velocity_env_cfg.py  # <- this is the base task configuration

The environments are then registered in the source/robot_lab/tasks/locomotion/velocity/config/unitree_a1/__init__.py:

gym.register(
    id="RobotLab-Isaac-Velocity-Flat-Unitree-A1-v0",
    entry_point="isaaclab.envs:ManagerBasedRLEnv",
    disable_env_checker=True,
    kwargs={
        "env_cfg_entry_point": f"{__name__}.flat_env_cfg:UnitreeA1FlatEnvCfg",
        "rsl_rl_cfg_entry_point": f"{agents.__name__}.rsl_rl_ppo_cfg:UnitreeA1FlatPPORunnerCfg",
    },
)

gym.register(
    id="RobotLab-Isaac-Velocity-Rough-Unitree-A1-v0",
    entry_point="isaaclab.envs:ManagerBasedRLEnv",
    disable_env_checker=True,
    kwargs={
        "env_cfg_entry_point": f"{__name__}.rough_env_cfg:UnitreeA1RoughEnvCfg",
        "rsl_rl_cfg_entry_point": f"{agents.__name__}.rsl_rl_ppo_cfg:UnitreeA1RoughPPORunnerCfg",
    },
)

Tensorboard

To view tensorboard, run:

tensorboard --logdir=logs

Code formatting

A pre-commit template is given to automatically format the code.

To install pre-commit:

pip install pre-commit

Then you can run pre-commit with:

pre-commit run --all-files

Troubleshooting

Pylance Missing Indexing of Extensions

In some VsCode versions, the indexing of part of the extensions is missing. In this case, add the path to your extension in .vscode/settings.json under the key "python.analysis.extraPaths".

Note: Replace <path-to-isaac-lab> with your own IsaacLab path.

{
    "python.analysis.extraPaths": [
        "${workspaceFolder}/source/robot_lab",
        "/<path-to-isaac-lab>/source/isaaclab",
        "/<path-to-isaac-lab>/source/isaaclab_assets",
        "/<path-to-isaac-lab>/source/isaaclab_mimic",
        "/<path-to-isaac-lab>/source/isaaclab_rl",
        "/<path-to-isaac-lab>/source/isaaclab_tasks",
    ]
}

Citation

Please cite the following if you use this code or parts of it:

@software{fan-ziqi2024robot_lab,
  author = {Ziqi Fan},
  title = {robot_lab: RL Extension Library for Robots, Based on IsaacLab.},
  url = {https://github.com/fan-ziqi/robot_lab},
  year = {2024}
}

Acknowledgements

The project uses some code from the following open-source code repositories:

Alescontrela/AMP_for_hardware

Name		Name	Last commit message	Last commit date
Latest commit History 94 Commits
.github		.github
.vscode		.vscode
docker		docker
scripts		scripts
source/robot_lab		source/robot_lab
.dockerignore		.dockerignore
.flake8		.flake8
.gitattributes		.gitattributes
.gitignore		.gitignore
.pre-commit-config.yaml		.pre-commit-config.yaml
CONTRIBUTORS.md		CONTRIBUTORS.md
LICENSE		LICENSE
README.md		README.md
pyproject.toml		pyproject.toml

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Repository files navigation

robot_lab

Overview

Installation

Docker setup

Building Isaac Lab Base Image

Building robot_lab Image

Running the container

Interacting with a running container

Shutting down the container

Try examples

Quadruped

Wheeled

Humanoid

AMP for Quadruped

HandStand

Add your own robot

Tensorboard

Code formatting

Troubleshooting

Pylance Missing Indexing of Extensions

Citation

Acknowledgements

About

Uh oh!

Releases

Packages

Languages

License

LegolasUndomiel/robot_lab

Folders and files

Latest commit

History

Repository files navigation

robot_lab

Overview

Installation

Docker setup

Building Isaac Lab Base Image

Building robot_lab Image

Running the container

Interacting with a running container

Shutting down the container

Try examples

Quadruped

Wheeled

Humanoid

AMP for Quadruped

HandStand

Add your own robot

Tensorboard

Code formatting

Troubleshooting

Pylance Missing Indexing of Extensions

Citation

Acknowledgements

About

Resources

License

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Languages

Packages