Isaac Gym: Overview & Getting Started

Learning Objectives

Set up our reinforcement learning example repository: OmniIsaacGymEnvs
1.Install OmniIsaacGymEnvs for Isaac Sim
2.Run inferencing and training examples in OmniIsaacGymEnvs
3.Install OmniIsaacGymEnvs in Docker
4.Run OmniIsaacGymEnvs examples with LiveStream

Getting started

Default Python Environment
Learn about Isaac Sim’s python environment and locate the python executable in Isaac Sim
Isaac Sim provides a built-in Python 3.10 environment that packages can use, similar to a system-level Python install
Recommended to use this Python environment when running the Python scripts

./python.sh path/to/script.py

Run the following from the Isaac Sim root folder to start a Python script in this environment
Workstation Installation
=> Already Done

Installing Examples Repository

(1)First clone the repository

git clone https://github.com/NVIDIA-Omniverse/OmniIsaacGymEnvs.git

(2)Install the examples as a python module in Isaac Sim
(2)-1 Locate the Isaac Sim python executable, which by default should be python.sh on Linux or python.bat on Windows, located at the root of the Isaac Sim directory

cd ~/.local/share/ov/pkg/isaac_sim-*/
ls -l python.sh
pwd

(2)-2 Refer to the Isaac Sim python executable path as PYTHON_PATH

alias PYTHON_PATH=~/.local/share/ov/pkg/isaac_sim-*/python.sh

(2)-3 Add it to .bashrc

nano ~/.bashrc
alias PYTHON_PATH='~/.local/share/ov/pkg/isaac_sim-*/python.sh'

(2)-4 Install OmniIsaacGymEnvs to PYTHON_PATH by running the following from the root of OmniIsaacGymEnvs

cd OmniIsaacGymEnvs/
PYTHON_PATH -m pip install -e .

=> The following error may appear during the initial installation. This error is harmless and can be ignored
(2)-5 List of current environments in OmniIsaacGymEnvs:

What I intend to test
=> Anymal & AnymalTerrain

Running Examples

Example scripts should be launched from OmniIsaacGymEnvs/omniisaacgymenvs

Launching Training Examples

Train your first policy

cd ~/OmniIsaacGymEnvs/omniisaacgymenvs
ls -l
PYTHON_PATH scripts/rlgames_train.py task=Cartpole

We will see an Isaac Sim window pop up
Once Isaac Sim initialization completes (which may take a few minutes if launching for the first time), the Cartpole scene will be constructed and simulation will start running automatically

The process will terminate once training finishes





The process will terminate once training finishes
Result
In the path /home/rosexpert1/OmniIsaacGymEnvs/omniisaacgymenvs/runs/, a Cartpole file is generated and the model with the best rewards is stored every 25, 50, 75, and 100 epochs
Anymal
Train your policy

cd ~/OmniIsaacGymEnvs/omniisaacgymenvs
PYTHON_PATH scripts/rlgames_train.py task=Anymal

We will see an Isaac Sim window pop up
Once Isaac Sim initialization completes (which may take a few minutes if launching for the first time), the Anymal scene will be constructed and simulation will start running automatically

Terminal





50 Epochs
79bdd6d86a)
100 Epochs

150 Epochs

200 Epochs

Checkpoint

250 Epochs & Checkpoint

300 Epochs

350 Epochs

400 Epochs

450 Epochs

500 Epochs

550 Epochs

600 Epochs

650 Epochs

700 Epochs

750 Epochs

800 Epochs

850 Epochs

900 Epochs

950 Epochs

1000 Epochs

Training Completed

Result

=> New Anymal file generated in the runs folder

=> Models generated per epochs is identified
Anymal Terrain
We will see an Isaac Sim window pop up
Once Isaac Sim initialization completes (which may take a few minutes if launching for the first time), the Anymal scene will be constructed and simulation will start running automatically

Terminal




50 Epochs

100 Epochs

Saving Checkpoint

150 Epochs

200 Epochs

250 Epochs

300 Epochs

350 Epochs

400 Epochs

450 Epochs

500 Epochs

550 Epochs

600 Epochs

650 Epochs

700 Epochs

750 Epochs

800 Epochs

850 Epochs

900 Epochs

950 Epochs

1000 Epochs

1050 Epochs

1100 Epochs

1150 Epochs

1200 Epochs

1250 Epochs

1300 Epochs

1350 Epochs

1400 Epochs

1450 Epochs

1500 Epochs

1550 Epochs

1600 Epochs

1650 Epochs

1700 Epochs

1750 Epochs

1800 Epochs

1850 Epochs

1900 Epochs

1950 Epochs

2000 Epochs

Training Completed

Result

cd ~/OmniIsaacGymEnvs/omniisaacgymenvs/runs/AnymalTerrain/nn/
ls -l

In the path /home/rosexpert1/OmniIsaacGymEnvs/omniisaacgymenvs/runs/, an AnymalTerrain file is generated and the model with the best rewards is stored every 25, 50, 75, and 100 epochs

Running Inference

cd ~/OmniIsaacGymEnvs/omniisaacgymenvs
PYTHON_PATH scripts/rlgames_train.py task=Cartpole test=True checkpoint=omniverse://localhost/NVIDIA/Assets/Isaac/2023.1.1/Isaac/Samples/OmniIsaacGymEnvs/Checkpoints/cartpole.pth

Error

Solution(Searching for it…)
https://forums.developer.nvidia.com/t/assistance-required-error-with-inferencing-using-pre-trained-checkpoints-in-isaac-gym/286386

Reference:
https://docs.omniverse.nvidia.com/isaacsim/latest/isaac_gym_tutorials/tutorial_gym_isaac_gym.html
https://github.com/NVIDIA-Omniverse/OmniIsaacGymEnvs/blob/main/docs/examples/rl_examples.md#anymal-anymalpy