Getting Started

We have provided a template to demonstrate how to use Isaac Lab. This template leverages a robotic arm and hand for reinforcement learning (RL) training. The template is open-sourced and available at: https://github.com/CyberNachos/isaacLab.manipulation/tree/main.

Installation and Deployment of IsaacLab.manipulation

• Clone repository and install:
  git clone https://github.com/CyberNachos/isaacLab.manipulation.git
  cd isaacLab.manipulation
  conda activate isaaclab
  pip install -e .
  

• Install RSL_RL in the isaacLab repository:
  cd isaacLab/manipulation/algorithms
  git clone https://github.com/leggedrobotics/rsl_rl.git
  cd rsl_rl
  python -m pip install -e .
  
  ## refresh index
  cd isaacLab.manipulation
  python -m pip install -e .
  

You can design your own RL Algorithm by editing "modules" and "algorithms" in RSL-RL

After completing the installation, you should be able to run the following examples:

python3 scripts/rsl_rl/train.py --task Template-Isaac-Reach-Kinova-v0--num_envs 4096 --headless
python3 scripts/rsl_rl/train.py --task Template-Isaac-Reach-Franka-v0 --num_envs 4096 --headless
python3 scripts/rsl_rl/train.py --task Template-Isaac-Reach-UR10-v0 --num_envs 4096 --headless
python3 scripts/rsl_rl/train.py --task Template-Isaac-Repose-Cube-Allegro-v0 --num_envs 4096 --headless

• The --task parameter specifies which task to execute. These tasks are registered in the environment using gym.register. You can use Ctrl+F to locate where they are registered or modify the task name during registration if needed.
• The --headless mode disables the graphical interface. This is particularly useful when running a large number of parallel environments (envs). If you accidentally enable it, you can turn off the graphical interface by pressing the V key. If you are running only a few environments, you may choose to keep the graphical interface enabled. It is generally recommended to keep the graphical interface on (by omitting --headless) during testing.