Quickstart Details

Quickstart Details#

Extended reference for topics introduced in Quickstart.

Quick Start Using Isaac Launchable#

For users first learning Isaac Lab without sufficient local compute, the Isaac Launchable project provides a browser-based Isaac Sim and Isaac Lab environment via NVIDIA Brev.

Click here to deploy

Running Tasks#

Use physics= and renderer= for backend selection and presets= for task-specific options (observation modes, camera configs, etc.). They fold into Hydra overrides automatically.

# Kit-less: Newton MJWarp + Newton visualizer
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/train.py \
  --task=Isaac-Cartpole-Direct-v0 \
  --num_envs=4096 \
  physics=newton_mjwarp --visualizer newton

# With Isaac Sim: PhysX
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/train.py \
  --task=Isaac-Cartpole-Direct-v0 \
  --num_envs=4096 \
  physics=physx

# Camera task: physics + renderer + domain preset
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/train.py \
  --task=Isaac-Cartpole-Camera-Presets-Direct-v0 \
  physics=newton_mjwarp renderer=newton_renderer presets=rgb

# OVRTX rendering (kit-less, no Kit visualizer)
./isaaclab.sh -p scripts/reinforcement_learning/rsl_rl/train.py \
  --task=Isaac-Repose-Cube-Shadow-Vision-Benchmark-Direct-v0 \
  --headless --enable_cameras --num_envs=16 --max_iterations=10 \
  physics=newton_mjwarp renderer=ovrtx_renderer presets=simple_shading_diffuse_mdl

isaaclab.bat -p scripts\reinforcement_learning\rsl_rl\train.py ^
  --task=Isaac-Cartpole-Direct-v0 ^
  --num_envs=4096 ^
  physics=newton_mjwarp --visualizer newton

Kit-less visualizer options: newton, rerun, viser. Combine with commas: --visualizer newton,rerun.

Available Presets#

Physics backends (physics=NAME):

  • physx — PhysX via Isaac Sim (default when no selector is given)

  • newton_mjwarp — Newton with the MuJoCo-Warp solver

  • newton_kamino — Newton with the Kamino solver (beta, limited tasks)

  • ovphysx — OV PhysX (kit-less; incompatible with --visualizer kit)

Renderer backends (renderer=NAME):

  • isaacsim_rtx_renderer — Isaac Sim RTX (default with Isaac Sim)

  • newton_renderer — Newton Warp renderer

  • ovrtx_renderer — OV RTX renderer (kit-less)

Domain presets (presets=NAME[,NAME,...]) are task-specific — run --task=<name> --help to list them.

Common combinations:

physics=newton_mjwarp renderer=newton_renderer presets=rgb
physics=newton_mjwarp renderer=newton_renderer presets=depth
physics=physx renderer=isaacsim_rtx_renderer presets=rgb
physics=physx renderer=isaacsim_rtx_renderer presets=depth
physics=physx renderer=isaacsim_rtx_renderer presets=albedo
physics=newton_mjwarp renderer=ovrtx_renderer presets=rgb
physics=newton_mjwarp renderer=ovrtx_renderer presets=simple_shading_diffuse_mdl

Legacy presets=newton_mjwarp,newton_renderer,rgb form still works; prefer typed selectors for clarity. See Hydra Configuration System for the full preset system.

List Available Environments#

Task names are registered with the Gymnasium API. List them with:

./isaaclab.sh -p scripts/environments/list_envs.py

Example output:

+--------+----------------------+--------------------------------------------+...
|   2    | Isaac-Ant-Direct-v0  |  isaaclab_tasks.direct.ant.ant_env:AntEnv  |...
|   48   | Isaac-Ant-v0         | isaaclab.envs:ManagerBasedRLEnv            |...

Each task may appear in Direct and ManagerBased variants — see Task Design Workflows for the two primary workflows.

Generate Your Own Project#

Scaffold a new project with the template generator:

./isaaclab.sh --new

Choose External vs Internal, Direct vs Manager, and RL Framework options, then install:

uv pip install -e source/<given-project-name>

The generated __init__.py registers the environment with Gymnasium:

gym.register(
    id="Template-isaaclabtutorial_env-v0",
    entry_point=f"{__name__}.isaaclabtutorial_env:IsaaclabtutorialEnv",
    disable_env_checker=True,
    kwargs={
        "env_cfg_entry_point": f"{__name__}.isaaclabtutorial_env_cfg:IsaaclabtutorialEnvCfg",
        "skrl_cfg_entry_point": f"{agents.__name__}.skrl_ppo_cfg:PPORunnerCfg",
    },
)

Configurations#

Configurations use the @configclass decorator and contain no __init__. Example from the cartpole environment:

@configclass
class CartpoleEnvCfg(DirectRLEnvCfg):
    decimation = 2
    episode_length_s = 5.0
    action_scale = 100.0
    action_space = 1
    observation_space = 4

    sim: SimulationCfg = SimulationCfg(dt=1 / 120, render_interval=decimation)
    robot_cfg: ArticulationCfg = CARTPOLE_CFG.replace(prim_path="/World/envs/env_.*/Robot")
    scene: InteractiveSceneCfg = InteractiveSceneCfg(num_envs=4096, env_spacing=4.0, replicate_physics=True)

    rew_scale_alive = 1.0
    rew_scale_terminated = -2.0

CLI arguments such as --num_envs override matching config fields at launch time.

Robots#

Robots are defined as configuration instances. See Robot Configurations for a full example.

Apps and Sims#

PhysX workflows require launching the Isaac Sim app. Newton workflows do not. For standalone scripts outside the standard task runners, use AppLauncher:

from isaaclab.app import AppLauncher

parser = argparse.ArgumentParser()
parser.add_argument("--num_envs", type=int, default=1)
AppLauncher.add_app_launcher_args(parser)
args_cli = parser.parse_args()

app_launcher = AppLauncher(args_cli)
simulation_app = app_launcher.app

Many Isaac Lab modules cannot be imported until the app is launched. See the Isaac Sim documentation for standalone app development.