Reinforcement Learning Scripts

Reinforcement Learning Scripts#

We provide wrappers to different reinforcement libraries. These wrappers convert the data from the environments into the respective libraries function argument and return types.

Preset Selectors#

All training and play scripts accept physics=NAME, renderer=NAME, and presets=NAME[,NAME,...] tokens appended directly to the command (no leading dashes). See Hydra Configuration System for all available names and how the selectors work.

# Switch physics backend
./isaaclab.sh -p scripts/reinforcement_learning/<framework>/train.py \
    --task <task-name> --headless physics=newton_mjwarp

# Switch renderer (camera environments)
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/train.py \
    --task Isaac-Cartpole-Camera-Presets-Direct-v0 --headless \
    --enable_cameras renderer=newton_renderer

# Combine selectors freely
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/train.py \
    --task Isaac-Cartpole-Camera-Presets-Direct-v0 --headless \
    --enable_cameras physics=newton_mjwarp renderer=newton_renderer presets=rgb

Note

Not all environments support the Newton backend yet. Using presets=newton_mjwarp with an environment that has not been configured for Newton will raise an error at launch. See Newton Backend for more details, and the Migrating to Isaac Lab 3.0 guide for how to add Newton support to your own environments.

Newton does not require Isaac Sim (kit-less mode). See Kit-less Installation for setup.

Observation-mode Presets#

Some environments support multiple observation modes selectable via presets=. Unlike physics or renderer presets, observation-mode presets affect the checkpoint structure: you must pass the same preset to both the training and play scripts. Using a different preset (or none) at play time will cause a model-architecture mismatch when loading the checkpoint.

For example, Isaac-Repose-Cube-Shadow-Vision-Direct-v0 defaults to RGB + depth + segmentation inputs but can be switched to RGB-only with presets=rgb:

# Train with RGB-only observations
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/train.py \
    --task Isaac-Repose-Cube-Shadow-Vision-Direct-v0 --headless \
    --enable_cameras presets=rgb

# Play — must use the same preset to load the matching checkpoint
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/play.py \
    --task Isaac-Repose-Cube-Shadow-Vision-Direct-Play-v0 \
    --enable_cameras presets=rgb

Other available presets for this environment: albedo, simple_shading_constant_diffuse, simple_shading_diffuse_mdl, simple_shading_full_mdl. The depth preset is intended for benchmarking only (see the environment’s config for details).

RL-Games#

Attention

When using RL-Games with the Ray workflow for distributed training or hyperparameter tuning, please be aware that due to security risks associated with Ray, this workflow is not intended for use outside of a strictly controlled network environment.

  • Training an agent with RL-Games on Isaac-Ant-v0:

    # install python module (for rl-games)
./isaaclab.sh -i rl_games
# run script for training
./isaaclab.sh -p scripts/reinforcement_learning/rl_games/train.py --task Isaac-Ant-v0 --headless
# run script for training with Newton backend
./isaaclab.sh -p scripts/reinforcement_learning/rl_games/train.py --task Isaac-Ant-v0 --headless presets=newton_mjwarp
# run script for playing with 32 environments
./isaaclab.sh -p scripts/reinforcement_learning/rl_games/play.py --task Isaac-Ant-v0 --num_envs 32 --checkpoint /PATH/TO/model.pth
# run script for recording video of a trained agent (requires installing `ffmpeg`)
./isaaclab.sh -p scripts/reinforcement_learning/rl_games/play.py --task Isaac-Ant-v0 --headless --video --video_length 200

    :: install python module (for rl-games)
isaaclab.bat -i rl_games
:: run script for training
isaaclab.bat -p scripts\reinforcement_learning\rl_games\train.py --task Isaac-Ant-v0 --headless
:: run script for training with Newton backend
isaaclab.bat -p scripts\reinforcement_learning\rl_games\train.py --task Isaac-Ant-v0 --headless presets=newton_mjwarp
:: run script for playing with 32 environments
isaaclab.bat -p scripts\reinforcement_learning\rl_games\play.py --task Isaac-Ant-v0 --num_envs 32 --checkpoint /PATH/TO/model.pth
:: run script for recording video of a trained agent (requires installing `ffmpeg`)
isaaclab.bat -p scripts\reinforcement_learning\rl_games\play.py --task Isaac-Ant-v0 --headless --video --video_length 200

RSL-RL#

  • Training an agent with RSL-RL on Isaac-Reach-Franka-v0:

    # install python module (for rsl-rl)
./isaaclab.sh -i rsl_rl
# run script for training
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/train.py --task Isaac-Reach-Franka-v0 --headless
# run script for training with Newton backend
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/train.py --task Isaac-Reach-Franka-v0 --headless presets=newton_mjwarp
# run script for playing with 32 environments
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/play.py --task Isaac-Reach-Franka-v0 --num_envs 32 --load_run run_folder_name --checkpoint /PATH/TO/model.pt
# run script for recording video of a trained agent (requires installing `ffmpeg`)
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/play.py --task Isaac-Reach-Franka-v0 --headless --video --video_length 200

    :: install python module (for rsl-rl)
isaaclab.bat -i rsl_rl
:: run script for training
isaaclab.bat -p scripts\reinforcement_learning\rsl_rl\train.py --task Isaac-Reach-Franka-v0 --headless
:: run script for training with Newton backend
isaaclab.bat -p scripts\reinforcement_learning\rsl_rl\train.py --task Isaac-Reach-Franka-v0 --headless presets=newton_mjwarp
:: run script for playing with 32 environments
isaaclab.bat -p scripts\reinforcement_learning\rsl_rl\play.py --task Isaac-Reach-Franka-v0 --num_envs 32 --load_run run_folder_name --checkpoint /PATH/TO/model.pt
:: run script for recording video of a trained agent (requires installing `ffmpeg`)
isaaclab.bat -p scripts\reinforcement_learning\rsl_rl\play.py --task Isaac-Reach-Franka-v0 --headless --video --video_length 200

  • Training and distilling an agent with RSL-RL on Isaac-Velocity-Flat-Anymal-D-v0:

    # install python module (for rsl-rl)
./isaaclab.sh -i rsl_rl
# run script for rl training of the teacher agent
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/train.py --task Isaac-Velocity-Flat-Anymal-D-v0 --headless
# run script for rl training of the teacher agent with Newton backend
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/train.py --task Isaac-Velocity-Flat-Anymal-D-v0 --headless presets=newton_mjwarp
# run script for distilling the teacher agent into a student agent
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/train.py --task Isaac-Velocity-Flat-Anymal-D-v0 --headless --agent rsl_rl_distillation_cfg_entry_point --load_run teacher_run_folder_name
# run script for playing the student with 64 environments
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/play.py --task Isaac-Velocity-Flat-Anymal-D-v0 --num_envs 64 --agent rsl_rl_distillation_cfg_entry_point

    :: install python module (for rsl-rl)
isaaclab.bat -i rsl_rl
:: run script for rl training of the teacher agent
isaaclab.bat -p scripts\reinforcement_learning\rsl_rl\train.py --task Isaac-Velocity-Flat-Anymal-D-v0 --headless
:: run script for rl training of the teacher agent with Newton backend
isaaclab.bat -p scripts\reinforcement_learning\rsl_rl\train.py --task Isaac-Velocity-Flat-Anymal-D-v0 --headless presets=newton_mjwarp
:: run script for distilling the teacher agent into a student agent
isaaclab.bat -p scripts\reinforcement_learning\rsl_rl\train.py --task Isaac-Velocity-Flat-Anymal-D-v0 --headless --agent rsl_rl_distillation_cfg_entry_point --load_run teacher_run_folder_name
:: run script for playing the student with 64 environments
isaaclab.bat -p scripts\reinforcement_learning\rsl_rl\play.py --task Isaac-Velocity-Flat-Anymal-D-v0 --num_envs 64 --agent rsl_rl_distillation_cfg_entry_point

SKRL#

  • Training an agent with SKRL on Isaac-Reach-Franka-v0:

    # install python module (for skrl)
./isaaclab.sh -i skrl
# run script for training
./isaaclab.sh -p scripts/reinforcement_learning/skrl/train.py --task Isaac-Reach-Franka-v0 --headless
# run script for training with Newton backend
./isaaclab.sh -p scripts/reinforcement_learning/skrl/train.py --task Isaac-Reach-Franka-v0 --headless presets=newton_mjwarp
# run script for playing with 32 environments
./isaaclab.sh -p scripts/reinforcement_learning/skrl/play.py --task Isaac-Reach-Franka-v0 --num_envs 32 --checkpoint /PATH/TO/model.pt
# run script for recording video of a trained agent (requires installing `ffmpeg`)
./isaaclab.sh -p scripts/reinforcement_learning/skrl/play.py --task Isaac-Reach-Franka-v0 --headless --video --video_length 200

    :: install python module (for skrl)
isaaclab.bat -i skrl
:: run script for training
isaaclab.bat -p scripts\reinforcement_learning\skrl\train.py --task Isaac-Reach-Franka-v0 --headless
:: run script for training with Newton backend
isaaclab.bat -p scripts\reinforcement_learning\skrl\train.py --task Isaac-Reach-Franka-v0 --headless presets=newton_mjwarp
:: run script for playing with 32 environments
isaaclab.bat -p scripts\reinforcement_learning\skrl\play.py --task Isaac-Reach-Franka-v0 --num_envs 32 --checkpoint /PATH/TO/model.pt
:: run script for recording video of a trained agent (requires installing `ffmpeg`)
isaaclab.bat -p scripts\reinforcement_learning\skrl\play.py --task Isaac-Reach-Franka-v0 --headless --video --video_length 200

    Warning

    It is recommended to install JAX manually before proceeding to install skrl and its dependencies, as JAX installs its CPU version by default. Visit the skrl installation page for more details. Note that JAX GPU support is only available on Linux.

    JAX 0.6.0 or higher (built on CuDNN v9.8) is incompatible with Isaac Lab’s PyTorch 2.7 (built on CuDNN v9.7), and therefore not supported. To install a compatible version of JAX for CUDA 12 use pip install "jax[cuda12]<0.6.0" "flax<0.10.7", for example.

    Hint

    When using JAX its default behavior is to pre-allocate 75% of the GPU memory for its own computations. If you run into memory issues, you can set the XLA_PYTHON_CLIENT_PREALLOCATE=false environment variable to disable this behavior, or reduce the amount of pre-allocated memory by setting export XLA_PYTHON_CLIENT_MEM_FRACTION=0.5 which will allocate 50% of the GPU memory for JAX. Any value between 0 and 1 can be set, where 0 will allocate no memory for JAX and 1 will allocate 100% of the GPU memory for JAX.

    # install python module (for skrl)
./isaaclab.sh -i skrl
# install jax<0.6.0 for torch 2.7
./isaaclab.sh -p -m pip install "jax[cuda12]<0.6.0" "flax<0.10.7"
# install skrl dependencies for JAX
./isaaclab.sh -p -m pip install skrl["jax"]
# run script for training
./isaaclab.sh -p scripts/reinforcement_learning/skrl/train.py --task Isaac-Reach-Franka-v0 --headless --ml_framework jax
# run script for training with Newton backend
./isaaclab.sh -p scripts/reinforcement_learning/skrl/train.py --task Isaac-Reach-Franka-v0 --headless --ml_framework jax presets=newton_mjwarp
# run script for playing with 32 environments
./isaaclab.sh -p scripts/reinforcement_learning/skrl/play.py --task Isaac-Reach-Franka-v0 --num_envs 32  --ml_framework jax --checkpoint /PATH/TO/model.pt
# run script for recording video of a trained agent (requires installing `ffmpeg`)
./isaaclab.sh -p scripts/reinforcement_learning/skrl/play.py --task Isaac-Reach-Franka-v0 --headless --ml_framework jax --video --video_length 200

    • Training the multi-agent environment Isaac-Shadow-Hand-Over-Direct-v0 with skrl:

    # install python module (for skrl)
./isaaclab.sh -i skrl
# run script for training with the MAPPO algorithm (IPPO is also supported)
./isaaclab.sh -p scripts/reinforcement_learning/skrl/train.py --task Isaac-Shadow-Hand-Over-Direct-v0 --headless --algorithm MAPPO
# run script for playing with 32 environments with the MAPPO algorithm (IPPO is also supported)
./isaaclab.sh -p scripts/reinforcement_learning/skrl/play.py --task Isaac-Shadow-Hand-Over-Direct-v0 --num_envs 32 --algorithm MAPPO --checkpoint /PATH/TO/model.pt

    :: install python module (for skrl)
isaaclab.bat -i skrl
:: run script for training with the MAPPO algorithm (IPPO is also supported)
isaaclab.bat -p scripts\reinforcement_learning\skrl\train.py --task Isaac-Shadow-Hand-Over-Direct-v0 --headless --algorithm MAPPO
:: run script for playing with 32 environments with the MAPPO algorithm (IPPO is also supported)
isaaclab.bat -p scripts\reinforcement_learning\skrl\play.py --task Isaac-Shadow-Hand-Over-Direct-v0 --num_envs 32 --algorithm MAPPO --checkpoint /PATH/TO/model.pt

Stable-Baselines3#

  • Training an agent with Stable-Baselines3 on Isaac-Velocity-Flat-Unitree-A1-v0:

    # install python module (for stable-baselines3)
./isaaclab.sh -i sb3
# run script for training
./isaaclab.sh -p scripts/reinforcement_learning/sb3/train.py --task Isaac-Velocity-Flat-Unitree-A1-v0 --headless
# run script for training with Newton backend
./isaaclab.sh -p scripts/reinforcement_learning/sb3/train.py --task Isaac-Velocity-Flat-Unitree-A1-v0 --headless presets=newton_mjwarp
# run script for playing with 32 environments
./isaaclab.sh -p scripts/reinforcement_learning/sb3/play.py --task Isaac-Velocity-Flat-Unitree-A1-v0 --num_envs 32 --checkpoint /PATH/TO/model.zip
# run script for recording video of a trained agent (requires installing `ffmpeg`)
./isaaclab.sh -p scripts/reinforcement_learning/sb3/play.py --task Isaac-Velocity-Flat-Unitree-A1-v0 --headless --video --video_length 200

    :: install python module (for stable-baselines3)
isaaclab.bat -i sb3
:: run script for training
isaaclab.bat -p scripts\reinforcement_learning\sb3\train.py --task Isaac-Velocity-Flat-Unitree-A1-v0 --headless
:: run script for training with Newton backend
isaaclab.bat -p scripts\reinforcement_learning\sb3\train.py --task Isaac-Velocity-Flat-Unitree-A1-v0 --headless presets=newton_mjwarp
:: run script for playing with 32 environments
isaaclab.bat -p scripts\reinforcement_learning\sb3\play.py --task Isaac-Velocity-Flat-Unitree-A1-v0 --num_envs 32 --checkpoint /PATH/TO/model.zip
:: run script for recording video of a trained agent (requires installing `ffmpeg`)
isaaclab.bat -p scripts\reinforcement_learning\sb3\play.py --task Isaac-Velocity-Flat-Unitree-A1-v0 --headless --video --video_length 200

RLinf#

RLinf is a distributed RL infrastructure for fine-tuning Vision-Language-Action (VLA) models such as GR00T. It uses Ray for distributed computing and FSDP for model parallelism, enabling RL training of large VLA models that don’t fit on a single GPU.

For installation instructions, see RL Post-Training for VLA Models.

  • Training a VLA agent with RLinf:

    # Train with a specific config
./isaaclab.sh -p scripts/reinforcement_learning/rlinf/train.py \
    --config_name isaaclab_ppo_gr00t_assemble_trocar \
    --model_path /path/to/checkpoint

  • Evaluating a trained VLA agent:

    # Evaluate with video recording
./isaaclab.sh -p scripts/reinforcement_learning/rlinf/play.py \
    --config_name isaaclab_ppo_gr00t_assemble_trocar \
    --model_path /path/to/checkpoint --video

All the scripts above log the training progress to Tensorboard in the logs directory in the root of the repository. The logs directory follows the pattern logs/<library>/<task>/<date-time>, where <library> is the name of the learning framework, <task> is the task name, and <date-time> is the timestamp at which the training script was executed.

To view the logs, run:

# execute from the root directory of the repository
./isaaclab.sh -p -m tensorboard.main --logdir=logs

:: execute from the root directory of the repository
isaaclab.bat -p -m tensorboard.main --logdir=logs