# isaac-sim/isaaclab architecture

> Architecture diagram of the GitHub repository isaac-sim/isaaclab, generated by GitDiagram from its file tree, README and a sample of source files (last updated 2026-09-25). It is an AI-made overview: check details against the source.

- Interactive diagram: https://gitdiagram.com/isaac-sim/isaaclab
- Repository: https://github.com/isaac-sim/isaaclab
- Video tour (about a minute): https://gitdiagram.com/isaac-sim/isaaclab/video

## Overview

Isaac Lab is a GPU-accelerated robotics simulation and learning framework built on Isaac Sim, with kit-less Newton workflows also supported. The principal workflow is to launch a training or evaluation script, step a configured environment, and use its observations and rewards to drive a learning framework. Environments coordinate scenes, physics, and task managers; separate subsystems provide sensors, control, teleoperation, imitation learning, and data tooling. The README establishes these major capabilities; arrows are limited to the environment lifecycle shown in sampled source and documented workflow relationships.

Main entry points: [`source/isaaclab/isaaclab/envs/manager_based_env.py`](https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab/isaaclab/envs/manager_based_env.py) (Environment API), [`source/isaaclab/isaaclab/physics/physics_manager.py`](https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab/isaaclab/physics/physics_manager.py) (Physics manager), [`scripts/reinforcement_learning/train.py`](https://github.com/isaac-sim/isaaclab/blob/develop/scripts/reinforcement_learning/train.py) (RL train), [`source/isaaclab/isaaclab/sim/simulation_context.py`](https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab/isaaclab/sim/simulation_context.py) (Simulation context)

## Diagram (Mermaid)

```mermaid
flowchart TD

subgraph group_workflow["Learning workflow"]
  node_train["RL train<br/>[train.py]"]
  node_play["RL play<br/>[play.py]"]
  node_imitation["Imitation scripts"]
  node_rl_frameworks["RL frameworks"]
end

subgraph group_environment["Environments"]
  node_env_api["Environment API"]
  node_env_types["RL and MARL envs"]
  node_task_managers["Task managers"]
  node_mdp_terms["MDP terms"]
end

subgraph group_simulation["Simulation"]
  node_scene["Interactive scene"]
  node_sim_context["Simulation context"]
  node_physics_api["Physics manager<br/>[physics_manager.py]"]
  node_physx["PhysX backend<br/>[physx_manager.py]"]
  node_newton["Newton backend<br/>[newton_manager.py]"]
  node_ovphysx["OVPhysX backend<br/>[ovphysx_manager.py]"]
  node_terrain["Terrain generation"]
end

subgraph group_robotics["Robot interaction"]
  node_sensors["Sensors"]
  node_renderers["Renderers"]
  node_controllers["Controllers"]
  node_actuators["Actuators"]
  node_teleop["Teleoperation devices"]
end

subgraph group_data["Data workflows"]
  node_demo_tools["Demo and dataset tools"]
  node_mimic["Mimic dataset workflow"]
  node_locomanipulation["Locomanipulation data"]
end

node_user(("Researcher"))
node_isaac_sim["Isaac Sim"]

node_user -->|"launches"| node_train
node_user -->|"launches"| node_play
node_user -->|"launches"| node_imitation
node_train -->|"steps environment"| node_env_api
node_play -->|"steps environment"| node_env_api
node_imitation -->|"uses environments"| node_env_api
node_train -->|"trains with"| node_rl_frameworks
node_env_api -->|"provides"| node_env_types
node_env_api -->|"creates scene"| node_scene
node_env_api -->|"controls lifecycle"| node_sim_context
node_env_api -->|"coordinates"| node_task_managers
node_task_managers -->|"evaluates terms"| node_mdp_terms
node_sim_context -->|"initializes"| node_physics_api
node_physics_api -->|"selects backend"| node_physx
node_physics_api -->|"selects backend"| node_newton
node_physics_api -->|"selects backend"| node_ovphysx
node_sim_context -->|"runs on"| node_isaac_sim
node_scene -->|"uses terrain"| node_terrain
node_env_api -->|"renders and observes"| node_sensors
node_sensors -->|"uses"| node_renderers
node_env_api -->|"accepts control"| node_controllers
node_controllers -->|"computes commands"| node_actuators
node_teleop -->|"provides input"| node_controllers
node_demo_tools -->|"records or replays"| node_env_api
node_mimic -->|"processes demonstrations"| node_demo_tools
node_locomanipulation -->|"generates datasets"| node_demo_tools

click node_train "https://github.com/isaac-sim/isaaclab/blob/develop/scripts/reinforcement_learning/train.py"
click node_play "https://github.com/isaac-sim/isaaclab/blob/develop/scripts/reinforcement_learning/play.py"
click node_imitation "https://github.com/isaac-sim/isaaclab/tree/develop/scripts/imitation_learning"
click node_env_api "https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab/isaaclab/envs/manager_based_env.py"
click node_env_types "https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/envs"
click node_task_managers "https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/managers"
click node_mdp_terms "https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/envs/mdp"
click node_scene "https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab/isaaclab/scene/interactive_scene.py"
click node_sim_context "https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab/isaaclab/sim/simulation_context.py"
click node_physics_api "https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab/isaaclab/physics/physics_manager.py"
click node_physx "https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab_physx/isaaclab_physx/physics/physx_manager.py"
click node_newton "https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab_newton/isaaclab_newton/physics/newton_manager.py"
click node_ovphysx "https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab_ov/isaaclab_ov/physics/ovphysx_manager.py"
click node_sensors "https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/sensors"
click node_renderers "https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/renderers"
click node_controllers "https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/controllers"
click node_actuators "https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/actuators"
click node_teleop "https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/devices"
click node_terrain "https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/terrains"
click node_demo_tools "https://github.com/isaac-sim/isaaclab/tree/develop/scripts/tools"
click node_mimic "https://github.com/isaac-sim/isaaclab/tree/develop/scripts/imitation_learning/isaaclab_mimic"
click node_locomanipulation "https://github.com/isaac-sim/isaaclab/tree/develop/scripts/imitation_learning/locomanipulation_sdg"

classDef toneNeutral fill:#f8fafc,stroke:#334155,stroke-width:1.5px,color:#0f172a
classDef toneBlue fill:#dbeafe,stroke:#2563eb,stroke-width:1.5px,color:#172554
classDef toneAmber fill:#fef3c7,stroke:#d97706,stroke-width:1.5px,color:#78350f
classDef toneMint fill:#dcfce7,stroke:#16a34a,stroke-width:1.5px,color:#14532d
classDef toneRose fill:#ffe4e6,stroke:#e11d48,stroke-width:1.5px,color:#881337
classDef toneIndigo fill:#e0e7ff,stroke:#4f46e5,stroke-width:1.5px,color:#312e81
classDef toneTeal fill:#ccfbf1,stroke:#0f766e,stroke-width:1.5px,color:#134e4a
class node_train,node_play,node_imitation,node_rl_frameworks toneBlue
class node_env_api,node_env_types,node_task_managers,node_mdp_terms toneAmber
class node_scene,node_sim_context,node_physics_api,node_physx,node_newton,node_ovphysx,node_terrain toneMint
class node_sensors,node_renderers,node_controllers,node_actuators,node_teleop toneRose
class node_demo_tools,node_mimic,node_locomanipulation,node_user,node_isaac_sim toneIndigo
```

## Components

### Learning workflow

- [RL train](https://github.com/isaac-sim/isaaclab/blob/develop/scripts/reinforcement_learning/train.py): `scripts/reinforcement_learning/train.py`
- [RL play](https://github.com/isaac-sim/isaaclab/blob/develop/scripts/reinforcement_learning/play.py): `scripts/reinforcement_learning/play.py`
- [Imitation scripts](https://github.com/isaac-sim/isaaclab/tree/develop/scripts/imitation_learning): `scripts/imitation_learning`
- **RL frameworks**

### Environments

- [Environment API](https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab/isaaclab/envs/manager_based_env.py): `source/isaaclab/isaaclab/envs/manager_based_env.py`
- [RL and MARL envs](https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/envs): `source/isaaclab/isaaclab/envs`
- [Task managers](https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/managers): `source/isaaclab/isaaclab/managers`
- [MDP terms](https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/envs/mdp): `source/isaaclab/isaaclab/envs/mdp`

### Simulation

- [Interactive scene](https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab/isaaclab/scene/interactive_scene.py): `source/isaaclab/isaaclab/scene/interactive_scene.py`
- [Simulation context](https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab/isaaclab/sim/simulation_context.py): `source/isaaclab/isaaclab/sim/simulation_context.py`
- [Physics manager](https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab/isaaclab/physics/physics_manager.py): `source/isaaclab/isaaclab/physics/physics_manager.py`
- [PhysX backend](https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab_physx/isaaclab_physx/physics/physx_manager.py): `source/isaaclab_physx/isaaclab_physx/physics/physx_manager.py`
- [Newton backend](https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab_newton/isaaclab_newton/physics/newton_manager.py): `source/isaaclab_newton/isaaclab_newton/physics/newton_manager.py`
- [OVPhysX backend](https://github.com/isaac-sim/isaaclab/blob/develop/source/isaaclab_ov/isaaclab_ov/physics/ovphysx_manager.py): `source/isaaclab_ov/isaaclab_ov/physics/ovphysx_manager.py`
- [Terrain generation](https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/terrains): `source/isaaclab/isaaclab/terrains`

### Robot interaction

- [Sensors](https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/sensors): `source/isaaclab/isaaclab/sensors`
- [Renderers](https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/renderers): `source/isaaclab/isaaclab/renderers`
- [Controllers](https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/controllers): `source/isaaclab/isaaclab/controllers`
- [Actuators](https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/actuators): `source/isaaclab/isaaclab/actuators`
- [Teleoperation devices](https://github.com/isaac-sim/isaaclab/tree/develop/source/isaaclab/isaaclab/devices): `source/isaaclab/isaaclab/devices`

### Data workflows

- [Demo and dataset tools](https://github.com/isaac-sim/isaaclab/tree/develop/scripts/tools): `scripts/tools`
- [Mimic dataset workflow](https://github.com/isaac-sim/isaaclab/tree/develop/scripts/imitation_learning/isaaclab_mimic): `scripts/imitation_learning/isaaclab_mimic`
- [Locomanipulation data](https://github.com/isaac-sim/isaaclab/tree/develop/scripts/imitation_learning/locomanipulation_sdg): `scripts/imitation_learning/locomanipulation_sdg`

### Other components

- **Researcher**
- **Isaac Sim**

## Connections

- Researcher → RL train: launches
- Researcher → RL play: launches
- Researcher → Imitation scripts: launches
- RL train → Environment API: steps environment
- RL play → Environment API: steps environment
- Imitation scripts → Environment API: uses environments
- RL train → RL frameworks: trains with
- Environment API → RL and MARL envs: provides
- Environment API → Interactive scene: creates scene
- Environment API → Simulation context: controls lifecycle
- Environment API → Task managers: coordinates
- Task managers → MDP terms: evaluates terms
- Simulation context → Physics manager: initializes
- Physics manager → PhysX backend: selects backend
- Physics manager → Newton backend: selects backend
- Physics manager → OVPhysX backend: selects backend
- Simulation context → Isaac Sim: runs on
- Interactive scene → Terrain generation: uses terrain
- Environment API → Sensors: renders and observes
- Sensors → Renderers: uses
- Environment API → Controllers: accepts control
- Controllers → Actuators: computes commands
- Teleoperation devices → Controllers: provides input
- Demo and dataset tools → Environment API: records or replays
- Mimic dataset workflow → Demo and dataset tools: processes demonstrations
- Locomanipulation data → Demo and dataset tools: generates datasets
