Metadata-Version: 2.4
Name: strands-robots
Version: 0.5.0
Summary: AI-powered robot control, simulation, and training for Strands Agents - integrates with MuJoCo, Isaac Sim, Newton, and many more
Project-URL: Homepage, https://github.com/strands-labs/robots
Project-URL: Documentation, https://github.com/strands-labs/robots#readme
Project-URL: Repository, https://github.com/strands-labs/robots.git
Project-URL: Issues, https://github.com/strands-labs/robots/issues
Project-URL: Changelog, https://github.com/strands-labs/robots/releases
Author-email: AWS <opensource@amazon.com>
License: Apache-2.0
License-File: LICENSE
License-File: NOTICE
Keywords: agents,gr00t,imitation-learning,isaac-sim,lerobot,manipulation,mujoco,reinforcement-learning,robot-learning,robots,sdk,simulation,strands,strands-agents,teleoperation,vla
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: Intended Audience :: Science/Research
Classifier: License :: OSI Approved :: Apache Software License
Classifier: Operating System :: OS Independent
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Classifier: Topic :: Scientific/Engineering :: Physics
Classifier: Topic :: Software Development :: Libraries :: Python Modules
Requires-Python: >=3.12
Requires-Dist: numpy<3.0.0,>=1.21.0
Requires-Dist: opencv-python-headless<5.0.0,>=4.5.0
Requires-Dist: pillow<13.0.0,>=8.0.0
Requires-Dist: strands-agents<2.0.0,>=1.0.0
Provides-Extra: all
Requires-Dist: adam-atan2-pytorch>=0.0.12; extra == 'all'
Requires-Dist: awscrt<1.0.0,>=0.20.0; extra == 'all'
Requires-Dist: awsiotsdk<2.0.0,>=1.21.0; extra == 'all'
Requires-Dist: boto3<2.0.0,>=1.34.0; extra == 'all'
Requires-Dist: device-connect-agent-tools>=0.1.0; extra == 'all'
Requires-Dist: device-connect-edge>=0.2.0; extra == 'all'
Requires-Dist: eclipse-zenoh<2.0.0,>=1.0.0; extra == 'all'
Requires-Dist: huggingface-hub<2.0.0,>=1.0; extra == 'all'
Requires-Dist: hydra-core<2.0.0,>=1.3.0; extra == 'all'
Requires-Dist: imageio-ffmpeg<1.0.0,>=0.4.0; extra == 'all'
Requires-Dist: imageio<3.0.0,>=2.28.0; extra == 'all'
Requires-Dist: json5<1.0.0,>=0.9.0; extra == 'all'
Requires-Dist: lerobot[async]<0.7.0,>=0.6.0; extra == 'all'
Requires-Dist: lerobot[dataset,feetech]<0.7.0,>=0.6.0; extra == 'all'
Requires-Dist: lerobot[molmoact2]<0.7.0,>=0.6.0; extra == 'all'
Requires-Dist: mink>=0.0.4; extra == 'all'
Requires-Dist: msgpack<2.0.0,>=1.0.0; extra == 'all'
Requires-Dist: mujoco<4.0.0,>=3.2.0; extra == 'all'
Requires-Dist: omegaconf<3.0.0,>=2.3.0; extra == 'all'
Requires-Dist: onnxruntime<2.0.0,>=1.17.0; extra == 'all'
Requires-Dist: pytorch-lightning<3.0.0,>=2.0.0; extra == 'all'
Requires-Dist: pyyaml<7.0.0,>=5.1; extra == 'all'
Requires-Dist: pyzmq<28.0.0,>=27.0.0; extra == 'all'
Requires-Dist: qpsolvers[daqp]>=4.0.0; extra == 'all'
Requires-Dist: robot-descriptions<2.0.0,>=1.11.0; extra == 'all'
Requires-Dist: roslibpy<2.0.0,>=1.7.0; extra == 'all'
Requires-Dist: scipy>=1.10.0; extra == 'all'
Requires-Dist: strands-agents[ollama]<2.0.0,>=1.0.0; extra == 'all'
Requires-Dist: torch>=2.0.0; extra == 'all'
Requires-Dist: vector-quantize-pytorch>=1.14.0; extra == 'all'
Requires-Dist: websockets>=12.0; extra == 'all'
Provides-Extra: benchmark-libero
Requires-Dist: libero<1.0.0,>=0.1.0; extra == 'benchmark-libero'
Provides-Extra: cosmos3-diffusers
Requires-Dist: accelerate>=0.26; extra == 'cosmos3-diffusers'
Requires-Dist: diffusers>=0.30; extra == 'cosmos3-diffusers'
Requires-Dist: torch>=2.0; extra == 'cosmos3-diffusers'
Requires-Dist: transformers>=4.40; extra == 'cosmos3-diffusers'
Provides-Extra: cosmos3-service
Requires-Dist: msgpack<2.0.0,>=1.0.0; extra == 'cosmos3-service'
Requires-Dist: websockets>=12.0; extra == 'cosmos3-service'
Provides-Extra: cosmos3-sim
Requires-Dist: mink>=0.0.4; extra == 'cosmos3-sim'
Requires-Dist: mujoco<4.0.0,>=3.2.0; extra == 'cosmos3-sim'
Requires-Dist: qpsolvers[quadprog]>=4.0.0; extra == 'cosmos3-sim'
Provides-Extra: curobo
Provides-Extra: dev
Requires-Dist: mypy<2.0.0,>=1.0.0; extra == 'dev'
Requires-Dist: pytest-cov<6.0.0,>=4.0.0; extra == 'dev'
Requires-Dist: pytest-timeout<3.0.0,>=2.0.0; extra == 'dev'
Requires-Dist: pytest<10.0.0,>=6.0; extra == 'dev'
Requires-Dist: ruff<0.16.0,>=0.15.12; extra == 'dev'
Provides-Extra: device-connect
Requires-Dist: device-connect-agent-tools>=0.1.0; extra == 'device-connect'
Requires-Dist: device-connect-edge>=0.2.0; extra == 'device-connect'
Provides-Extra: groot-service
Requires-Dist: msgpack<2.0.0,>=1.0.0; extra == 'groot-service'
Requires-Dist: pyzmq<28.0.0,>=27.0.0; extra == 'groot-service'
Provides-Extra: inference
Requires-Dist: websockets>=12.0; extra == 'inference'
Provides-Extra: lerobot
Requires-Dist: lerobot[dataset,feetech]<0.7.0,>=0.6.0; extra == 'lerobot'
Provides-Extra: lerobot-async
Requires-Dist: lerobot[async]<0.7.0,>=0.6.0; extra == 'lerobot-async'
Requires-Dist: lerobot[dataset,feetech]<0.7.0,>=0.6.0; extra == 'lerobot-async'
Provides-Extra: mesh
Requires-Dist: eclipse-zenoh<2.0.0,>=1.0.0; extra == 'mesh'
Requires-Dist: json5<1.0.0,>=0.9.0; extra == 'mesh'
Provides-Extra: mesh-iot
Requires-Dist: awscrt<1.0.0,>=0.20.0; extra == 'mesh-iot'
Requires-Dist: awsiotsdk<2.0.0,>=1.21.0; extra == 'mesh-iot'
Requires-Dist: boto3<2.0.0,>=1.34.0; extra == 'mesh-iot'
Requires-Dist: eclipse-zenoh<2.0.0,>=1.0.0; extra == 'mesh-iot'
Requires-Dist: json5<1.0.0,>=0.9.0; extra == 'mesh-iot'
Provides-Extra: molmoact2
Requires-Dist: lerobot[dataset,feetech]<0.7.0,>=0.6.0; extra == 'molmoact2'
Requires-Dist: lerobot[molmoact2]<0.7.0,>=0.6.0; extra == 'molmoact2'
Provides-Extra: motionbricks
Requires-Dist: adam-atan2-pytorch>=0.0.12; extra == 'motionbricks'
Requires-Dist: hydra-core<2.0.0,>=1.3.0; extra == 'motionbricks'
Requires-Dist: omegaconf<3.0.0,>=2.3.0; extra == 'motionbricks'
Requires-Dist: pytorch-lightning<3.0.0,>=2.0.0; extra == 'motionbricks'
Requires-Dist: scipy>=1.10.0; extra == 'motionbricks'
Requires-Dist: torch>=2.0.0; extra == 'motionbricks'
Requires-Dist: vector-quantize-pytorch>=1.14.0; extra == 'motionbricks'
Provides-Extra: moveit2
Requires-Dist: msgpack<2.0.0,>=1.0.0; extra == 'moveit2'
Requires-Dist: pyzmq<28.0.0,>=27.0.0; extra == 'moveit2'
Provides-Extra: ollama
Requires-Dist: strands-agents[ollama]<2.0.0,>=1.0.0; extra == 'ollama'
Provides-Extra: ros2
Requires-Dist: cyclonedds<1.0.0,>=0.10.2; extra == 'ros2'
Provides-Extra: rosbridge
Requires-Dist: roslibpy<2.0.0,>=1.7.0; extra == 'rosbridge'
Provides-Extra: sim
Requires-Dist: robot-descriptions<2.0.0,>=1.11.0; extra == 'sim'
Provides-Extra: sim-gs
Requires-Dist: gsplat<2.0.0,>=1.4.0; extra == 'sim-gs'
Requires-Dist: plyfile<2.0.0,>=1.0.0; extra == 'sim-gs'
Requires-Dist: torch<3.0.0,>=2.2.0; extra == 'sim-gs'
Provides-Extra: sim-isaac
Requires-Dist: imageio-ffmpeg<1.0.0,>=0.4.0; extra == 'sim-isaac'
Requires-Dist: imageio<3.0.0,>=2.28.0; extra == 'sim-isaac'
Requires-Dist: robot-descriptions<2.0.0,>=1.11.0; extra == 'sim-isaac'
Requires-Dist: usd-core<27.0.0,>=25.5; extra == 'sim-isaac'
Provides-Extra: sim-mujoco
Requires-Dist: imageio-ffmpeg<1.0.0,>=0.4.0; extra == 'sim-mujoco'
Requires-Dist: imageio<3.0.0,>=2.28.0; extra == 'sim-mujoco'
Requires-Dist: mink>=0.0.4; extra == 'sim-mujoco'
Requires-Dist: mujoco<4.0.0,>=3.2.0; extra == 'sim-mujoco'
Requires-Dist: qpsolvers[daqp]>=4.0.0; extra == 'sim-mujoco'
Requires-Dist: robot-descriptions<2.0.0,>=1.11.0; extra == 'sim-mujoco'
Provides-Extra: sim-newton
Requires-Dist: imageio-ffmpeg<1.0.0,>=0.4.0; extra == 'sim-newton'
Requires-Dist: imageio<3.0.0,>=2.28.0; extra == 'sim-newton'
Requires-Dist: mink>=0.0.4; extra == 'sim-newton'
Requires-Dist: mujoco-warp>=3.8.0; extra == 'sim-newton'
Requires-Dist: mujoco<4.0.0,>=3.2.0; extra == 'sim-newton'
Requires-Dist: newton<2.0.0,>=1.3.0; extra == 'sim-newton'
Requires-Dist: qpsolvers[daqp]>=4.0.0; extra == 'sim-newton'
Requires-Dist: robot-descriptions<2.0.0,>=1.11.0; extra == 'sim-newton'
Requires-Dist: trimesh<5.0.0,>=4.0.0; extra == 'sim-newton'
Requires-Dist: warp-lang<2.0.0,>=1.14.0; extra == 'sim-newton'
Provides-Extra: vera-sim
Requires-Dist: gym-pusht==0.1.5; extra == 'vera-sim'
Requires-Dist: gymnasium==0.29.1; extra == 'vera-sim'
Requires-Dist: imageio; extra == 'vera-sim'
Requires-Dist: mujoco>=3.5.0; extra == 'vera-sim'
Requires-Dist: robosuite==1.4.1; extra == 'vera-sim'
Provides-Extra: wbc
Requires-Dist: huggingface-hub<2.0.0,>=1.0; extra == 'wbc'
Requires-Dist: onnxruntime<2.0.0,>=1.17.0; extra == 'wbc'
Requires-Dist: pyyaml<7.0.0,>=5.1; extra == 'wbc'
Description-Content-Type: text/markdown

<div align="center">
  <div>
    <a href="https://strandsagents.com">
      <img src="https://strandsagents.com/latest/assets/logo-github.svg" alt="Strands Agents" width="55px" height="105px">
    </a>
  </div>

  <h1>
    Strands Robots
  </h1>

  <h2>
    Control, simulate, and train robots with natural language
  </h2>

  <div align="center">
    <a href="https://pypi.org/project/strands-robots/"><img alt="PyPI Version" src="https://img.shields.io/pypi/v/strands-robots"/></a>
    <a href="https://github.com/strands-labs/robots"><img alt="GitHub stars" src="https://img.shields.io/github/stars/strands-labs/robots"/></a>
    <a href="https://github.com/strands-labs/robots/blob/main/LICENSE"><img alt="License" src="https://img.shields.io/github/license/strands-labs/robots"/></a>
    <a href="https://github.com/google-deepmind/mujoco"><img alt="MuJoCo" src="https://img.shields.io/badge/MuJoCo-3.x-000000"/></a>
    <a href="https://github.com/NVIDIA/Isaac-GR00T"><img alt="GR00T" src="https://img.shields.io/badge/NVIDIA-GR00T-76B900?logo=nvidia"/></a>
    <a href="https://github.com/huggingface/lerobot"><img alt="LeRobot" src="https://img.shields.io/badge/🤗-LeRobot-yellow"/></a>
  </div>

  <p>
    <a href="https://strandsagents.com/">Strands Docs</a>
    ◆ <a href="https://github.com/google-deepmind/mujoco">MuJoCo</a>
    ◆ <a href="https://github.com/NVIDIA/Isaac-GR00T">NVIDIA GR00T</a>
    ◆ <a href="https://github.com/huggingface/lerobot">LeRobot</a>
    ◆ <a href="https://github.com/strands-labs/robots-sim">Robots Sim</a>
    ◆ <a href="https://github.com/orgs/strands-labs/projects/2">Project Board</a>
  </p>
</div>

<p align="center">
  <img src="docs/assets/hero_loop.svg" alt="Strands Robots - perceive, reason, act, world: the closed control loop around a Strands Agent core" width="100%">
</p>

`strands-robots` gives a [Strands Agent](https://github.com/strands-agents/harness-sdk)
hands. One `Robot()` call returns a **MuJoCo simulation** (default - no GPU, no
hardware) or a **real robot** - same code, same natural-language control, both
auto-joined to a peer-to-peer **mesh**.

```python
from strands import Agent
from strands_robots import Robot

robot = Robot("so100")              # MuJoCo sim by default; mode="real" for hardware
Agent(tools=[robot])("pick up the red cube")
```

### One agent, the whole robotics loop

Teleoperate a real arm to collect demos, fine-tune a policy on them, run it in
sim **and** on hardware, hand work to a fleet peer, and expose it all on ROS 2 -
one library, one mental model. Every line below is a distinct capability:

```python
from strands import Agent
from strands_robots import Robot
from strands_robots.tools import train_policy

# 1. TELEOPERATE a real SO-101 with its leader arm and RECORD demos as a
#    LeRobotDataset (one prompt drives cameras + teleop + recording).
follower = Robot("so101", mode="real", port="/dev/ttyACM0",
                 cameras={"front": {"type": "opencv", "index_or_path": "/dev/video0"}})
follower.attach_teleop("so101_leader", port="/dev/ttyACM1", id="leader")
Agent(tools=[follower])(
    "start_recording(repo_id='me/pick', root='/tmp/pick', fps=30, "
    "task='pick up the cube'); teleoperate for 60s; stop_recording"
)

# 2. POST-TUNE a policy on those demos (LoRA fine-tune; GPU box).
train_policy(action="train", provider="lerobot_local",
             dataset_root="/tmp/pick", base_model="lerobot/smolvla_base",
             output_dir="/tmp/pick_ckpt", method="lora", steps=20000)

# 3. RUN the tuned checkpoint - same policy on a MuJoCo twin AND the real arm.
twin = Robot("so101")                                              # sim twin, no hardware
twin.run_policy(robot_name="so101", policy_provider="lerobot_local",
                policy_config={"pretrained_name_or_path": "/tmp/pick_ckpt"}, duration=10.0)
follower.start_task("pick up the cube", policy_provider="lerobot_local",
                    policy_port=None, duration=10.0)               # real arm, in-process

# 4. COORDINATE a fleet - tell a mesh peer to assist, in natural language.
follower.mesh.tell(follower.mesh.peers[0]["peer_id"], "hold the tray steady")

# 5. EXPOSE the running sim on ROS 2 - rviz / nav2 / any ros2 node can subscribe.
from strands_robots.simulation import Simulation
sim = Simulation(ros2_bridge=True); sim.create_world(); sim.add_robot("so101")
sim.step(100)   # publishes /so101/joint_states + camera image_raw on the ROS 2 graph
```

| Step | Capability | Surface |
|------|------------|---------|
| 1 | Teleop + dataset recording | `Robot(mode="real")`, `attach_teleop`, `start_recording` |
| 2 | Policy post-tuning | `train_policy` (LeRobot / GR00T trainers) |
| 3 | Sim + hardware policy rollout | `run_policy` (sim), `start_task` (hardware) |
| 4 | Fleet coordination | `robot.mesh.tell` / `robot_mesh` tool |
| 5 | ROS 2 interop | `Simulation(ros2_bridge=True)`, `use_ros` |

> Steps 1 and 3-real need hardware; step 2 needs a GPU. Everything runs in sim
> with no hardware (`Robot("so101")`), so you can exercise the whole loop today.

## Why strands-robots

- **Sim-first, safe by default.** `Robot("so100")` spins up a MuJoCo world. You
  never accidentally drive real servos - `mode="real"` is an explicit opt-in.
- **70+ robots, 8 categories.** Arms, humanoids, quadrupeds, hands, drones,
  bimanual rigs - resolved from a single registry with auto-download of assets.
- **Any policy.** VLA models (NVIDIA GR00T, LeRobot ACT/Pi0/SmolVLA/Diffusion),
  plus classical motion planners, MPC, and scripted controllers behind one ABC.
- **Mesh networking built in.** Every robot is a Zenoh peer. `tell()` another
  robot what to do; broadcast an E-STOP; bridge to AWS IoT Core for fleets.
- **67-action simulation tool.** World building, physics, rendering, domain
  randomization, procedural terrain (`create_world(terrain="rough"|"stairs"|"pyramid"|"slope")`
  for locomotion), and LeRobotDataset recording - all agent-callable.
- **ROS 2 interop.** Observe + command any ROS 2 graph (`use_ros`), act as a
  robot with no rclpy (`use_rtps`), or expose a running sim as a ROS node.
- **One mental model.** Sim and hardware share the same policy interface,
  the same mesh, and the same natural-language control surface.

## How it works

<p align="center">
  <img src="docs/assets/architecture_flow.svg" alt="Strands Robots architecture - four-layer stack (Agent, Policies, Backends, Robots) with action signals flowing down and observation signals flowing back up" width="100%">
</p>

```mermaid
graph LR
    A[Natural Language<br/>'Pick up the red block'] --> B[Strands Agent]
    B --> C[Robot<br/>sim or real]
    C --> D[Policy Provider<br/>GR00T / Cosmos 3 / LeRobot / planner / mock]
    D --> E[Action Chunk]
    E --> F[MuJoCo Sim<br/>or Hardware]
    F -->|observation| C

    classDef input fill:#2ea44f,stroke:#1b7735,color:#fff
    classDef agent fill:#0969da,stroke:#044289,color:#fff
    classDef policy fill:#8250df,stroke:#5a32a3,color:#fff
    classDef hardware fill:#bf8700,stroke:#875e00,color:#fff

    class A input
    class B,C agent
    class D,E policy
    class F hardware
```

## Installation

Examples use [`uv`](https://docs.astral.sh/uv/) (`curl -LsSf https://astral.sh/uv/install.sh | sh`); plain `pip` works too.

```bash
uv pip install strands-robots
```

The base install is light (numpy, opencv-headless, Pillow). Pull in only the
extras you need:

| Extra | Installs | Use for |
|-------|----------|---------|
| `sim-mujoco` | MuJoCo, robot_descriptions, imageio, mink + qpsolvers[daqp] | Simulation (recommended starting point). mink/qpsolvers are the differential-IK solver behind the `move_to` Cartesian transport primitive; `qpsolvers` ships no solver of its own, so the `[daqp]` backend extra is declared with it. |
| `sim-newton` | Newton, Warp, MuJoCo-Warp, trimesh | GPU-native simulation (NVIDIA GPU; batched envs, headless ray-traced render) |
| `sim-isaac` | usd-core, imageio (Isaac Sim installed separately) | NVIDIA Isaac Sim backend - photorealistic RTX rendering, synthetic data, GPU-batched sensors, USD-native scenes. Install Isaac Sim itself separately: via its pip wheels on Python 3.12 (`isaacsim[all,extscache]` from pypi.nvidia.com - see the caveats in [`docs/simulation/isaac.md`](docs/simulation/isaac.md)), the Omniverse Launcher, Isaac Lab, or the NGC docker image. This extra pulls only the pip-installable Python helpers. (NVIDIA RTX GPU; GPU-only, not in `[all]`.) |
| `sim-gs` | gsplat, plyfile, torch | 3D Gaussian Splatting hybrid rendering (`strands_robots.rendering`): composite any sim backend's robot over a captured photoreal 3DGS scene. `gsplat` ships as a source dist that JIT-compiles CUDA kernels via `nvcc` on first use - probe with `strands_robots.rendering.gsplat_rasterizer_available()`; the zero-GPU `PanoramaBackground` works without this extra. (CUDA GPU; GPU-only, not in `[all]`.) |
| `lerobot` | LeRobot | Real hardware, local VLA inference, dataset recording |
| `molmoact2` | LeRobot + transformers, peft, scipy | MolmoAct2 transformers-native VLA (resolves from PyPI via lerobot >= 0.6) |
| `groot-service` | pyzmq, msgpack | NVIDIA GR00T inference client |
| `cosmos3-service` | websockets, msgpack | NVIDIA Cosmos 3 policy-server client |
| `curobo` | _(empty; install cuRobo from source)_ | In-process collision-aware motion planning (CUDA GPU) |
| `wbc` | onnxruntime | GR00T Whole-Body-Control (SONIC) humanoid locomotion - in-process ONNX, no GPU |
| `motionbricks` | torch + vector-quantize-pytorch, pytorch-lightning, hydra-core (install `motionbricks` from source) | NVIDIA MotionBricks generative kinematic motion for the G1 - in-process torch, composes with `wbc` |
| `mesh` | eclipse-zenoh, json5 | Peer-to-peer robot mesh |
| `mesh-iot` | awsiotsdk, awscrt, boto3 | AWS IoT Core mesh transport for fleets |
| `device-connect` | device-connect-edge, device-connect-agent-tools | Device-aware networking - discovery, RPC, events, safety (falls back to the built-in mesh if absent) |
| `benchmark-libero` | libero | LIBERO benchmark evaluation |
| `all` | everything above except the GPU-only `sim-isaac` / `sim-gs` extras | Kitchen sink |

```bash
# Most users start here:
uv pip install "strands-robots[sim-mujoco]"

# Real hardware + local policies:
uv pip install "strands-robots[sim-mujoco,lerobot]"

# MolmoAct2 VLA (transformers-native; resolves from PyPI via lerobot >= 0.6):
uv pip install "strands-robots[molmoact2]"

# Everything:
uv pip install "strands-robots[all]"
```

The **Isaac Sim** GPU backend is a built-in, in-tree peer of `mujoco` and
`newton` (it lives at `strands_robots.simulation.isaac`). Its pip-installable
helpers ship in the `sim-isaac` extra, but the Isaac Sim runtime itself (~30 GB)
is provisioned separately - via its own pip wheels on Python 3.12
(`pip install 'isaacsim[all,extscache]==6.0.*' --extra-index-url https://pypi.nvidia.com`,
with coverage/EULA caveats documented in
[`docs/simulation/isaac.md`](docs/simulation/isaac.md)), the Omniverse
Launcher, Isaac Lab, or the NGC docker image. Install the helpers with
`pip install 'strands-robots[sim-isaac]'`, then select the backend with
`create_simulation("isaac")` - see
[Simulation (MuJoCo)](#simulation-mujoco) and
[`docs/simulation/isaac.md`](docs/simulation/isaac.md).

From source:

```bash
git clone https://github.com/strands-labs/robots
cd robots
uv pip install -e ".[all,dev]"
```

## Quick starts

### Simulation (no GPU, no hardware)

```python
from strands import Agent
from strands_robots import Robot

robot = Robot("so100") # MuJoCo simulation
agent = Agent(tools=[robot])
agent("Wave the arm using the mock policy for 200 steps, then render a top-down view")
```

`Robot("so100")` returns a `Simulation` instance - the full 67-action
simulation AgentTool. Drive it in natural language through an `Agent`, call its
methods directly (`robot.render(camera_name="topdown")`), or dispatch an action
by calling it (`robot(action="render", camera_name="topdown")`). See
[Simulation](#simulation-mujoco).

> **Note:** `Robot("so100")` already creates the world **and** adds the robot
> for you. Do **not** call `create_world()` again on the returned instance -
> it will error with *"World already exists."* The `create_world()` /
> `add_robot()` sequence shown in [Simulation (MuJoCo)](#simulation-mujoco) is
> for the low-level `Simulation(...)` constructor, which starts empty.

### Real hardware + GR00T

```python
from strands import Agent
from strands_robots import Robot, gr00t_inference

robot = Robot(
    "so101",
    mode="real",
    cameras={
        "front": {"type": "opencv", "index_or_path": "/dev/video0", "fps": 30},
        "wrist": {"type": "opencv", "index_or_path": "/dev/video2", "fps": 30},
    },
    port="/dev/ttyACM0",
    data_config="so100_dualcam",
)

agent = Agent(tools=[robot, gr00t_inference])

# Start the GR00T inference service (Docker, Jetson/x86 GPU)
agent.tool.gr00t_inference(
    action="start",
    checkpoint_path="/data/checkpoints/model",
    port=8000,
    data_config="so100_dualcam",
)

agent("Use so101 to pick up the red block with the GR00T policy on port 8000")
```

### Local LeRobot policy (no inference server)

```python
from strands_robots import create_policy

# Direct HuggingFace inference - ACT, Pi0, SmolVLA, Diffusion, ...
policy = create_policy("lerobot/act_aloha_sim_transfer_cube_human")
```

### Teleoperation (leader arms, gamepads, WASD)

Drive any real robot - or a simulation - from one or more LeRobot
teleoperators. `Teleoperator()` mirrors the `Robot()` factory; `attach_teleop()`
+ `teleoperate()` run the control loop.

```python
from strands_robots import Robot, Teleoperator

# Leader arm -> follower arm (both speak {motor}.pos -> zero config)
follower = Robot("so101", mode="real", port="/dev/ttyACM0")
follower.attach_teleop("so101_leader", port="/dev/ttyACM1", id="leader")
follower.teleoperate()                       # Ctrl+C or stop_teleoperate()

# Earth Rover Mini+ with WASD keys (velocity keys -> zero config)
rover = Robot("earthrover_mini_plus", mode="real", robot_ip="192.168.1.151")
rover.attach_teleop("keyboard_rover")        # W/A/S/D
rover.teleoperate(block=True, duration=30)

# Cross-vocabulary or sim teleop -> supply a map_fn(action) -> action
robot.attach_teleop("keyboard_ee", map_fn=my_ik)   # EE deltas -> joint .pos
robot.teleoperate(publish=True)              # also stream over the mesh
```

17 teleoperators (`so100/so101/koch/omx/openarm` leaders, `bi_*` leaders,
`gamepad`, `keyboard`, `keyboard_ee`, `keyboard_rover`, `phone`,
`reachy2_teleoperator`, `unitree_g1`, homunculus arm/glove) drive 14 robots.
Zero-config when action keys match; otherwise pass `map_fn`. Full matrix +
recipes: [Teleoperation docs](https://strands-labs.github.io/robots/hardware/teleoperation/).

## Recording & streaming datasets

The physical-AI data loop, end to end: **record** a LeRobotDataset from sim or
hardware, **stream** it straight back for eval/training (no full download), and
optionally **dump** it to a mutable Hugging Face Storage Bucket. Needs the
`lerobot` extra (which bundles `datasets` + `av` + `torchcodec`).

```python
from strands import Agent
from strands_robots import Robot

sim = Robot("so100", mesh=False)
agent = Agent(tools=[sim])

# 1. COLLECT — one natural-language prompt drives scene + cameras + policy + record.
agent(
    "Create a world with the so100 robot, add a red cube and a front camera, "
    "start recording (repo_id='local/demo', root='/tmp/demo', fps=30, "
    "overwrite=True, task='pick up the red cube'), run the mock policy for "
    "60 steps, then stop recording."
)

# 2. STREAM — read it back lazily; camera frames decode on the fly from the MP4
#    shards, state/action from parquet. Nothing is re-materialized to disk.
reader = sim.stream_dataset("local/demo", root="/tmp/demo", shuffle=False)
for frame in reader:
    frame["observation.images.front"]   # (3, H, W) tensor, decoded from video
    frame["observation.state"]          # joint vector
    frame["action"]
    break
```

`stream_dataset()` is the in-process read counterpart to
`start_recording`/`stop_recording`. For full training, the upstream trainer uses
the same engine — `lerobot-train --policy.type=act --dataset.repo_id=... --dataset.streaming=true --num_workers=4`
(the `lerobot-train` entry point wraps `python -m lerobot.scripts.lerobot_train`;
flags are draccus `--dotted.key=value` form).

**Verify episode integrity.** A recording's ground truth is the parquet under
`meta/episodes/`, not the count a model narrates while collecting. Collect
episodes with a deterministic Python loop (one `run_policy(..., n_episodes=1)`
plus `save_episode()` per episode) rather than trusting a model to count its own
tool calls, then confirm the dataset holds the episodes you intended - in-process
or from the shell:

```python
sim.verify_dataset_episodes(expected=20)   # reads parquet; status="error" on a mega-episode
```

```bash
# exit 0 = pass, 1 = fail, so it drops straight into CI as a dataset gate
strands-robots verify-dataset /tmp/demo --expected 20
```

This catches the "mega-episode" corruption class - a run that buffered every
frame into one `episode_index=0` episode while reporting `20/20` - plus
`meta/info.json` vs parquet drift and zero-length episodes.

**Dump to a Storage Bucket** during collection (mutable, Xet-deduplicated — the
Phase 1/2 collection target that avoids git-LFS history bloat) with one kwarg:

```python
sim.stop_recording(bucket="your-org/robot-fave")   # → hf://buckets/your-org/robot-fave/demo
```

Requires the `hf` CLI with the `buckets`/`sync` subcommands
(`pip install -U "huggingface_hub>=1.0"` + `hf auth login` — 0.x releases of
`huggingface_hub` ship an `hf` entry point without them).

Any on-disk dataset directory can be synced (or daily re-synced) without a live
recording session — one recorded earlier in the process, or on hardware via
`lerobot-record`:

```python
from strands_robots import sync_dataset_to_bucket

sync_dataset_to_bucket("/tmp/demo", "your-org/robot-fave")
# → {"status": "success", "bucket_uri": "hf://buckets/your-org/robot-fave/demo"}
```

`run_id` defaults to the directory name; pass `run_id="nightly"` to choose the
bucket subpath, and `delete=True` for mirror semantics.

**Proprio-only / no video** (e.g. edge devices without a torchcodec wheel):
`sim.stream_dataset(repo_id, drop_videos=True, delta_timestamps={...})` streams
state/action only and never touches the video decoder. `drop_videos=True`
requires a `delta_timestamps` with at least one non-video key (e.g.
`{"observation.state": [0.0], "action": [0.0]}`) - without one, every feature
including video would stream, so the call raises `ValueError` instead of
silently no-opping.

> **macOS note (zero-touch).** torchcodec links ffmpeg via `@rpath`, and
> Homebrew's ffmpeg (`/opt/homebrew/lib`) is not on the default dyld search
> path — so video decode would normally fail with
> `Library not loaded: @rpath/libavutil.NN.dylib`. On `import strands_robots`
> we auto-detect this and put Homebrew's ffmpeg on `DYLD_FALLBACK_LIBRARY_PATH`
> (re-exec'ing the interpreter once for a plain script run; never inside
> Jupyter/REPL/pytest, where it just prints the one-line `export` to run). It's
> a no-op off macOS, without torchcodec, or when the var is already set. Disable
> with `STRANDS_ROBOTS_NO_DYLD_SHIM=1`. See `examples/06_agent_collect_and_stream.py`.

See also [Recording & datasets](docs/recording.md) for the `DatasetRecorder`
direct API and append/resume workflow.

## The `Robot()` factory

`Robot()` is a factory, not a wrapper - you get the real backend instance back
with all its methods.

```python
Robot("so100")                       # mode="sim"  (default, safe)
Robot("so100", mode="real")          # explicit hardware opt-in
Robot("so100", mode="auto")          # probe USB for servos, fall back to sim
Robot("my_arm", urdf_path="arm.xml") # bring your own MJCF/URDF
```

| Parameter | Type | Default | Description |
|-----------|------|---------|-------------|
| `name` | `str` | required | Robot name or alias (see [Supported robots](#supported-robots)) |
| `mode` | `str` | `"sim"` | `"sim"`, `"real"`, or `"auto"` (case-insensitive) |
| `backend` | `str` | `"mujoco"` | Sim backend: `"mujoco"`, `"newton"`, or `"isaac"` (all built-in; `isaac` needs the `sim-isaac` extra) |
| `urdf_path` | `str` | `None` | Explicit MJCF/URDF path (skips registry lookup) |
| `cameras` | `dict` | `None` | Camera config (**`mode="real"` only**) |
| `position` | `list[float]` | `[0,0,0]` | Spawn position in the sim world |
| `data_config` | `str` | name | Observation/action schema name |
| `mesh` | `bool` | `True` | Auto-join the Zenoh mesh |

Safety/validation rules:
- **Defaults to sim.** Real hardware is always an explicit `mode="real"`.
- **`cameras=` is rejected in sim mode** - add sim cameras via the `add_camera`
  action after creation.
- **Unknown robot names raise `ValueError`** unless you pass `urdf_path=`.
- **`STRANDS_ROBOT_MODE`** overrides detection; a typo'd value logs a warning
  and falls back to sim.

## Supported robots

70+ robots across 8 categories, resolved from
[`registry/robots.json`](strands_robots/registry/robots.json). Assets
(MJCF + meshes) auto-download from
[robot_descriptions](https://github.com/robot-descriptions/robot_descriptions.py)
/ [MuJoCo Menagerie](https://github.com/google-deepmind/mujoco_menagerie) on
first use. List them at runtime with `from strands_robots import list_robots; list_robots()`.

| Category | Count | Robots |
|----------|-------|--------|
| **Arm** | 22 | so100, so101, koch, omx, panda, fr3, fr3_v2, ur5e, ur10e, xarm7, kinova_gen3, kuka_iiwa, sawyer, piper, yam, z1, vx300s, wx250s, arx_l5, openarm, hope_jr, dynamixel_2r |
| **Humanoid** | 18 | unitree_g1, unitree_h1, unitree_h1_2, apollo, talos, reachy2, rby1, fourier_n1, booster_t1, adam_lite, asimov_v0, cassie, elf2, jvrc, op3, open_duck_mini, toddlerbot_2xc, toddlerbot_2xm |
| **Mobile** | 13 | spot, go1, unitree_go2, unitree_a1, aliengo, anymal_b, anymal_c, stretch, stretch3, lekiwi, tiago_dual, earthrover, robot_soccer_kit |
| **Hand** | 8 | shadow_hand, shadow_dexee, allegro_hand, leap_hand, ability_hand, aero_hand, robotiq_2f85, robotiq_2f85_v4 |
| **Bimanual** | 3 | aloha, bi_openarm, trossen_wxai |
| **Aerial** | 2 | crazyflie, skydio_x2 |
| **Expressive** | 1 | reachy_mini |
| **Mobile manip** | 1 | google_robot |

**Hardware-capable** (drivable with `mode="real"` via LeRobot): `so100`,
`so101`, `koch`, `omx`, `hope_jr`, `aloha`, `bi_openarm`, `reachy2`,
`unitree_g1`, `lekiwi`, `earthrover`. All are simulatable.

### Adding a robot

There are two paths, depending on whether the robot needs project-specific
metadata:

1. **Standard `robot_descriptions` robot (zero config).** Any MJCF robot shipped
   by [robot_descriptions](https://github.com/robot-descriptions/robot_descriptions.py)
   resolves automatically without a `robots.json` entry - the asset is
   discovered and downloaded on first use:

   ```python
   from strands_robots import Robot, list_discoverable

   sim = Robot("iiwa14")          # discovered, not in robots.json
   print(list_discoverable())     # the MJCF long tail you can load directly
   ```

   A curated `robots.json` entry always wins over discovery, so overriding a
   discovered robot later is non-breaking.

2. **Custom or metadata-rich robot.** If the robot needs a non-default joint
   count, hardware port, aliases, scene tweaks, or local mesh overrides, add a
   curated entry. For a robot that belongs in the shipped catalog, add it to
   [`registry/robots.json`](strands_robots/registry/robots.json) and open a PR.
   For a machine-local robot, register it at runtime instead of editing the
   package:

   ```python
   from strands_robots.registry import register_robot

   register_robot(name="my_arm", model_xml="my_arm.xml",
                  asset_dir="~/robots/my_arm", joints=7, category="arm")
   ```

## Tools reference

Import any of these and pass to `Agent(tools=[...])`. Each is a Strands
AgentTool returning `{"status", "content"}`.

| Tool | Purpose |
|------|---------|
| `Robot(...)` | Universal robot - sim or hardware, natural-language + async control |
| `run_policy` | Multi-episode policy rollout with per-episode eval + dataset recording |
| `train_policy` | Post-tune (fine-tune) a policy on a recorded dataset (LeRobot / GR00T trainers, full or LoRA) |
| `use_lerobot` | Universal LeRobot bridge - call ANY lerobot module/class/config directly (like `use_aws` wraps boto3) |
| `lerobot_train` | Thin local wrapper over the `lerobot-train` CLI (the engine behind `train_policy`) |
| `robot_mesh` | Coordinate robots over the Zenoh mesh (`tell`, `broadcast`, E-STOP) |
| `use_ros` | Bridge to any ROS 2 graph - list/echo/publish topics, call services (in-process rclpy) |
| `use_rtps` | Join a ROS 2 graph as a DDS participant - publish/echo topics, act as a robot (pure cyclonedds, no rclpy, all ROS 2 distros) |
| `gr00t_inference` | Manage NVIDIA GR00T inference services (Docker lifecycle) |
| `lerobot_camera` | OpenCV / RealSense camera discovery, capture, record |
| `lerobot_calibrate` | List, view, back up, restore LeRobot calibrations |
| `lerobot_teleoperate` | Record demonstrations, replay episodes |
| `pose_tool` | Store, recall, and execute named robot poses |
| `harness_memory` | Persist task solution traces + global success rules / failure models across agent sessions (Harness-VLA-style memory) |
| `serial_tool` | Low-level Feetech servo / raw serial communication |
| `download_assets` | Pre-fetch robot MJCF + meshes into the asset cache |

<details>
<summary><b>Robot tool actions</b></summary>

| Action | Parameters | Description |
|--------|------------|-------------|
| `execute` | `instruction`, `policy_port`, `duration` | Blocking execution until complete |
| `start` | `instruction`, `policy_port`, `duration` | Non-blocking async start |
| `status` | - | Current task status |
| `stop` | - | Interrupt running task (emergency stop) |
In sim mode the same tool exposes the 67 Simulation actions - see Simulation (MuJoCo).
</details>

<details>
<summary><b>GR00T inference tool actions</b></summary>

| Action | Parameters | Description |
|--------|------------|-------------|
| `start` | `checkpoint_path`, `port`, `data_config` | Start inference service |
| `stop` | `port` | Stop service on port |
| `status` | `port` | Check service status |
| `list` | - | List running services |
| `find_containers` | - | Find GR00T Docker containers |
| `build_image` / `download_checkpoint` / `start_container` | - | Full container lifecycle orchestration |

**TensorRT** acceleration:

```python
agent.tool.gr00t_inference(
    action="start",
    checkpoint_path="/data/checkpoints/model",
    port=8000,
    use_tensorrt=True,
    vit_dtype="fp8",     # ViT:  fp16 | fp8
    llm_dtype="nvfp4",   # LLM:  fp16 | nvfp4 | fp8
    dit_dtype="fp8",     # DiT:  fp16 | fp8
)
```

</details>

<details>
<summary><b>Camera / serial / pose / teleop tool actions</b></summary>

**Camera** - `discover`, `capture`, `capture_batch`, `record`, `preview`, `test`
**Serial** - `list_ports`, `feetech_position`, `feetech_ping`, `send`, `monitor`
**Pose** - `store_pose`, `load_pose`, `list_poses`, `move_motor`, `incremental_move`, `reset_to_home`
**Teleop** - `start`, `stop`, `list`, `replay`

</details>

## Policy providers

All policies implement one ABC - `async get_actions(observation, instruction, **kwargs)`.
The interface is deliberately agnostic about *how* actions are produced, so it
fits both VLA models and classical controllers.

```python
from strands_robots import create_policy

create_policy("mock")                                  # sinusoidal test actions
create_policy("groot", port=5555)                      # NVIDIA GR00T via ZMQ
create_policy("zmq://localhost:5555")                  # same, by URL
create_policy("cosmos3", embodiment="droid", port=8000)  # NVIDIA Cosmos 3 via WebSocket
create_policy("lerobot/act_aloha_sim_transfer_cube")   # local HF inference
```

| Provider | Backend | Notes |
|----------|---------|-------|
| `mock` | none | Sinusoidal trajectories; `requires_images=False` (~10x faster) |
| `groot` | NVIDIA GR00T N1.5/N1.6/N1.7 | Service mode (ZMQ to a Docker container) or local in-process (`model_path=`) |
| `cosmos3` | NVIDIA Cosmos 3 omnimodal VLA | Service mode (WebSocket to a Cosmos Framework RoboLab policy server); embodiments: `droid`, `umi`, `av`, `bridge` |
| `lerobot_local` | HuggingFace | Direct ACT / Pi0 / SmolVLA / Diffusion inference, no server |
| `lerobot_async` | HuggingFace via gRPC | Offload a LeRobot policy to a remote `PolicyServer` over lerobot's native async-inference gRPC transport (edge/light robot host) |
| `remote` | any policy, over WebSocket | Drop-in client that forwards observations to a remote `PolicyServer` and returns its action chunk: `create_policy("remote", endpoint="ws://gpu-box:8765")` (or the smart string `create_policy("ws://gpu-box:8765")`). For a light robot host with a GPU box elsewhere; mirrors the server policy's RTC support |
| `vera` | MIT VERA (DFoT/WAN planner + Jacobian IDM) | Two-stage video-to-action over a WebSocket GPU server (Docker); PushT + MimicGen, IK for eef-delta arms. **Git-only** (not on PyPI, no extra): `pip install 'vera @ git+https://github.com/sizhe-li/VERA.git'` plus `websockets msgpack numpy` |

```mermaid
classDiagram
    class Policy {
        <<abstract>>
        +get_actions(obs, instruction, **kwargs)
        +set_robot_state_keys(keys)
        +requires_images
        +reset(seed)
        +provider_name
    }
    class Gr00tPolicy
    class Cosmos3Policy
    class LerobotLocalPolicy
    class MockPolicy
    class YourPolicy
    Policy <|-- Gr00tPolicy
    Policy <|-- Cosmos3Policy
    Policy <|-- LerobotLocalPolicy
    Policy <|-- MockPolicy
    Policy <|-- YourPolicy
```

<details>
<summary><b>GR00T data configs (embodiment schemas)</b></summary>

A `data_config` defines the video + state keys GR00T expects for an
embodiment. 27 ship in
[`policies/groot/data_configs.json`](strands_robots/policies/groot/data_configs.json);
the common ones:

| Config | Cameras | Description |
|--------|---------|-------------|
| `so100` / `so101` | 1 (`video.webcam`) | Single-arm, single camera |
| `so100_dualcam` / `so101_dualcam` | 2 (front + wrist) | Single-arm, dual camera |
| `so100_4cam` | 4 (front, wrist, top, side) | Single-arm, quad camera |
| `so101_tricam` | 3 (front, wrist, side) | Single-arm, tri camera |
| `fourier_gr1_arms_only` | 1 (ego) | Fourier GR-1 bimanual arms + hands |
| `unitree_g1` | 1 (ego) | G1 upper body (arms + hands) |
| `unitree_g1_full_body` / `_locomanip` | - | G1 legs + waist + arms + hands |
| `bimanual_panda_gripper` | 3 | Dual Franka, EEF pose + gripper |
| `libero_panda` | 2 (image + wrist) | LIBERO benchmark Panda |
| `oxe_droid` / `oxe_google` / `oxe_widowx` | 1-2 | Open X-Embodiment schemas |
| `agibot_*` / `galaxea_r1_pro` | 3 | AgiBot / Galaxea humanoids |

Pick the config matching your robot's camera + state layout; pass it as
`data_config=` to `Robot(...)`, `gr00t_inference(...)`, or `create_policy("groot", ...)`.

</details>

> **Security:** `lerobot_local` loads HuggingFace models with
> `trust_remote_code=True` (arbitrary code execution). You must opt in with
> `export STRANDS_TRUST_REMOTE_CODE=1`. Only load models you trust.

### Cosmos 3 (NVIDIA omnimodal VLA - service mode)

[`nvidia/Cosmos3-Nano-Policy-DROID`](https://huggingface.co/nvidia/Cosmos3-Nano-Policy-DROID) via a self-contained WebSocket client (`cosmos3` / `c3` / `cosmos3://host:port`); no `openpi-client` dep, no `numpy<2` pin, so it composes with `lerobot` in one env.

<details>
<summary><b>Cosmos 3 server + client setup, embodiments, sim rollout</b></summary>

[`nvidia/Cosmos3-Nano-Policy-DROID`](https://huggingface.co/nvidia/Cosmos3-Nano-Policy-DROID)
served by the Cosmos Framework RoboLab WebSocket policy server. The policy
client is **self-contained** - it speaks the server's msgpack+NumPy wire
protocol directly via `websockets` + a vendored numpy packer (no
`openpi-client` dependency, no `numpy<2` pin), so it composes cleanly with
`lerobot` for dataset recording in the same env.

**1. Start the server** (holds the GPU), from a Cosmos Framework checkout:

```bash
uv sync --all-extras --group=cu130-train --group=policy-server
python -m cosmos_framework.scripts.action_policy_server_robolab \
    --checkpoint-path nvidia/Cosmos3-Nano-Policy-DROID --port 8000
curl http://localhost:8000/healthz   # -> 200 when ready (~4 min cold)
```

**2. Install the client** (the `cosmos3-service` extra ships only `msgpack`
+ `websockets` - numpy-version agnostic):

```bash
uv pip install -e '.[sim-mujoco]'
uv pip install 'strands-robots[cosmos3-service]'
```

**3. Use it** (`cosmos3`, `c3`, `cosmos3://host:port`, or the HF model-id all
resolve to `Cosmos3Policy`):

```python
from strands_robots.policies import create_policy

policy = create_policy("cosmos3", embodiment="droid", port=8000)
policy.set_robot_state_keys([f"joint_{i}" for i in range(7)] + ["gripper"])
chunk = policy.get_actions_sync(observation, "pick up the cube")
# chunk == [{"joint_0": .., ..., "gripper": ..}, ...]  (one dict per timestep)
```

The `droid` embodiment (`joint_pos`/RoboArena) conditions on **all three**
camera views and the server rejects a partial observation. Your
`observation_mapping` must map a sim/robot camera onto each of
`observation/wrist_image_left`, `observation/exterior_image_1_left`, and
`observation/exterior_image_2_left`; an incomplete mapping raises an actionable
client-side `ValueError` naming the missing keys before any request is sent
(other embodiments such as `umi`/`av`/`bridge` need only `observation/image`):

```python
policy = create_policy(
    "cosmos3", embodiment="droid", port=8000,
    observation_mapping={
        "wrist":     "observation/wrist_image_left",
        "exterior":  "observation/exterior_image_1_left",
        "exterior2": "observation/exterior_image_2_left",
    },
)
```

**4. Roll out in MuJoCo** - the `droid` embodiment drives a Franka/DROID-class
arm, so use the `franka` (or `panda`) sim asset:

```bash
MUJOCO_GL=egl python examples/vla/cosmos3_sim_rollout.py --record /tmp/c3.mp4
```

Embodiments: `droid` (10D, chunk 32, 15 fps), `umi`, `av`, `bridge`. If the
server is not running, the policy raises a `ConnectionError` with the exact
command to start it.


</details>

### Non-VLA policies (motion planners, MPC, scripted)

The same interface fits cuRobo, MoveIt2, OMPL, MPC, and pure-IK / scripted
trajectories - anything mapping `(observation, goal)` to joint targets.
Non-VLA providers set `requires_images = False` (skip camera rendering) and
read their goal from **well-known `**kwargs` keys** instead of parsing the
instruction string:

| Key | Type | Meaning |
|-----|------|---------|
| `target_pose` | `list[float]` | Cartesian goal `[x, y, z, qw, qx, qy, qz]` in base frame |
| `target_joints` | `dict[str, float]` | Joint-space goal keyed by joint name (rad / m) |
| `world_update` | `dict \| None` | Per-call world refresh for collision-aware planners |

Providers MUST ignore unknown `**kwargs` rather than raising, so callers can
pass shared keys across providers without coupling to a backend.

```python
from typing import Any
from strands_robots.policies import Policy, register_policy, create_policy


class ReachPolicy(Policy):
    """Linear interpolation from current joint state to target_joints."""

    def __init__(self, steps: int = 32, **_: Any) -> None:
        self._keys: list[str] = []
        self._steps = steps

    @property
    def provider_name(self) -> str:
        return "reach"

    @property
    def requires_images(self) -> bool:
        return False  # joint-state only -- skip camera rendering

    def set_robot_state_keys(self, robot_state_keys: list[str]) -> None:
        self._keys = list(robot_state_keys)

    async def get_actions(self, observation_dict, instruction, **kwargs):
        target = kwargs.get("target_joints")
        if target is None:
            raise ValueError("ReachPolicy requires target_joints kwarg")
        state = observation_dict.get("observation.state", [0.0] * len(self._keys))
        out = []
        for s in range(1, self._steps + 1):
            alpha = s / self._steps
            out.append({k: (1 - alpha) * state[i] + alpha * target[k]
                        for i, k in enumerate(self._keys)})
        return out


register_policy("reach", lambda: ReachPolicy, aliases=["lerp"])
policy = create_policy("reach")
```

<details>
<summary><b>Reference non-VLA providers: MoveIt2, cuRobo, WBC/SONIC</b></summary>

Three reference implementations of the goal-kwarg contract above. Each has a
runnable example + full install/deploy notes in its linked doc:

| Provider | Alias | Runs | Goal kwarg | Needs | Docs |
|----------|-------|------|-----------|-------|------|
| `moveit2` | `moveit` | ZMQ sidecar (ROS 2 / `moveit_py`, out-of-process) | `target_pose` / `target_joints` | `[moveit2]` extra (`pyzmq`, `msgpack`); a running sidecar | [MoveIt2 docs](https://strands-labs.github.io/robots/policies/moveit2/) |
| `curobo` | `cumotion` | in-process CUDA | `target_pose` / `target_joints` (+ `world_update`) | NVIDIA GPU; cuRobo from source (not on PyPI) | [cuRobo source](https://github.com/NVlabs/curobo) |
| `wbc` | `sonic` | in-process ONNX (CPU) | `target_velocity` `[vx, vy, omega]` | `[wbc]` extra (`onnxruntime`); a SONIC checkpoint | [WBC docs](https://strands-labs.github.io/robots/policies/wbc/) |
| `motionbricks` | `motion_bricks` | in-process torch (CPU/CUDA) | `style` / `mode`, `target_velocity`, `target_heading` | `[motionbricks]` extra + `motionbricks` from source + git-LFS checkpoints | [MotionBricks docs](https://strands-labs.github.io/robots/policies/motionbricks/) |

```python
from strands_robots.policies import create_policy

# Collision-aware planning (GPU, in-process); plan is cached, streamed per tick.
policy = create_policy("curobo", robot_config="franka.yml", action_horizon=16)
actions = policy.get_actions_sync(
    {"observation.state": [0.0, -0.79, 0.0, -2.36, 0.0, 1.57, 0.79]},
    "reach for the red block",                  # ignored by planners
    target_pose=[0.5, 0.0, 0.4, 1.0, 0.0, 0.0, 0.0],
)
```

Agents share one goal vocabulary across VLA and planner providers:
`Robot.start_task(..., policy_provider="curobo", target_pose=[...])` and
`mesh.tell(peer, "...", policy_provider="curobo", target_pose=[...])` flow the
same `target_pose` / `target_joints` / `world_update` kwargs through.

</details>


## Training providers

`create_trainer` is the training-side peer of `create_policy`: the provider
name is the same one used for inference, so `create_policy("groot")` and
`create_trainer("groot")` refer to one provider. It returns a `Trainer` you
drive with a `TrainSpec`, and the `train_policy` tool wraps this same factory
for agent-callable post-tuning.

```python
from strands_robots.training import create_trainer, TrainSpec

# Imitation / post-tuning on a recorded LeRobotDataset:
trainer = create_trainer("lerobot_local")
trainer.train(TrainSpec(dataset_root="/tmp/pick", base_model="lerobot/smolvla_base",
                        output_dir="/tmp/pick_ckpt", method="lora", steps=20000))

# From-scratch reinforcement learning in sim (no dataset):
ppo = create_trainer("ppo")   # or create_trainer("fast_sac")
```

| Provider | Kind | Notes |
|----------|------|-------|
| `lerobot_local` | Imitation / post-tuning | LeRobot fine-tune (full or LoRA) on a LeRobotDataset |
| `groot` | Imitation / post-tuning | NVIDIA GR00T fine-tune; needs an `embodiment` tag |
| `cosmos3` | Imitation / post-tuning | NVIDIA Cosmos 3 fine-tune (multi-node HSDP capable) |
| `mock` | Imitation (test) | No-op trainer for tests and dry runs |
| `ppo` | Reinforcement learning | On-policy PPO; pairs with `VecSimEnv` for parallel rollouts |
| `fast_sac` | Reinforcement learning | Off-policy Soft Actor-Critic |

The RL trainers (`ppo`, `fast_sac`) subclass `BaseRLAlgo` and share the same
`validate -> prepare -> train -> export` lifecycle as the imitation trainers.
They collect trajectories through `VecSimEnv` (N independent `SimEnv` as one
batched env) and score with `BaseRLAlgo.evaluate()`. The training package stays
torch-free until an RL provider is resolved on first use.


## Simulation (MuJoCo)

`Robot("so100")` (sim mode) returns a `Simulation` - a MuJoCo-backed AgentTool
exposing **67 actions** for world composition, physics, rendering, policy
execution, and dataset recording. Build it directly when you want full control:

```python
from strands_robots.simulation import Simulation

sim = Simulation(tool_name="sim", mesh=False)
sim.create_world()
sim.add_robot(name="arm", data_config="so100")
sim.add_object(name="cube", shape="box", position=[0.3, 0, 0.05])
sim.add_camera(name="topdown", position=[0, 0, 1.5], target=[0, 0, 0])

# Wrist camera: mount ON the gripper body so it tracks the arm like the real
# SO101/SO100 hardware cam. position/target are in the body's LOCAL frame.
# Body names are namespaced "<robot>/<body>" (e.g. "arm/gripper").
sim.add_camera(name="wrist", position=[0, -0.05, 0], target=[0, -0.15, 0],
               parent_body="arm/gripper")

sim.run_policy(robot_name="arm", policy_provider="mock", n_steps=200,
               control_frequency=50.0)

frame = sim.render(camera_name="topdown")   # {status, content:[text, image]}
```

<details>
<summary><b>The actions, grouped</b></summary>

- **World & scene**: `create_world`, `load_scene`, `replace_scene_mjcf`,
  `patch_scene_mjcf`, `reset`, `get_state`, `save_state`, `load_state`,
  `destroy`, `export_xml`.
- **Robots**: `add_robot`, `remove_robot`, `list_robots`, `get_robot_state`,
  `list_urdfs`, `register_urdf`, `get_features`.
- **Objects**: `add_object`, `remove_object`, `move_object`, `list_objects`.
- **Cameras & rendering**: `add_camera`, `remove_camera`, `render`,
  `render_depth`, `render_all`, `get_world_point`, `start_cameras_recording`,
  `stop_cameras_recording`, `get_cameras_recording_status`.
- **Physics**: `step`, `set_timestep`, `set_gravity`, `apply_force`, `raycast`,
  `multi_raycast`, `get_contacts`, `get_contact_forces`, `get_body_state`,
  `set_joint_positions`, `set_joint_velocities`, `forward_kinematics`,
  `get_jacobian`, `get_mass_matrix`, `inverse_dynamics`, `get_total_mass`,
  `get_energy`, `get_sensor_data`, `set_body_properties`, `set_geom_properties`.
- **Policy**: `run_policy`, `start_policy`, `stop_policy`,
  `list_policies_running`, `replay_episode`, `eval_policy`.
- **Randomization**: `randomize`.
- **Recording (LeRobotDataset)**: `start_recording`, `stop_recording`,
  `get_recording_status`.
- **Benchmarks**: `list_benchmarks`, `register_benchmark_from_file`,
  `evaluate_benchmark`.
- **Viewer**: `open_viewer`, `close_viewer`.

</details>

<details>
<summary><b>Common footguns</b></summary>

- **Planes must be static.** `add_object(shape="plane")` auto-sets
  `is_static=True`; passing `is_static=False` is a hard error.
- **Aim cameras.** Pass `target=[x,y,z]` to look at a point; `target == position`
  errors.
- **Wrist cameras mount on a body.** Pass `parent_body="<robot>/gripper"` to
  `add_camera` so the camera rides with the arm (realistic SO101/SO100 wrist
  cam). In that mode `position`/`target` are in the body's LOCAL frame, not
  world coordinates. Omit `parent_body` for a world-fixed camera.
- **MP4 vs dataset recording.** `start_cameras_recording` writes plain MP4
  (`[sim-mujoco]` only). `start_recording` writes a LeRobotDataset (parquet +
  MP4 + schema) and needs the `[lerobot]` extra.
- **Policy running → mutations blocked.** While a policy runs, state-mutating
  actions error with *"Cannot 'X' while a policy is running."* Stop it first.
- **Horizon parameters.** `run_policy` takes either `duration` or `n_steps`
  (both with `control_frequency`). `fast_mode=True` skips the between-step
  sleep for batch eval / data collection.
- **Name collisions.** Objects, bodies, robots, and cameras share the MuJoCo
  name table. Multi-robot joints/actuators are namespaced `{robot}/{joint}`.

</details>

**Self-healing:** unknown parameters are rejected with *"Unknown parameter X
for action Y. Valid: [...]"*, missing required params produce *"Action X
requires parameter Y."*, and vectors/dtypes are validated before MuJoCo sees
them - so the agent learns the contract without crashing the process.

**Third-party backends.** `create_simulation(name)` discovers backends beyond
the built-in `mujoco`/`newton`/`isaac` registry via Python
[entry points](https://packaging.python.org/en/latest/specifications/entry-points/).
A sibling package registers its `SimEngine` subclasses under the
`strands_robots.backends` group in its `pyproject.toml`, and they become
available on `pip install` without patching this package:

```toml
[project.entry-points."strands_robots.backends"]
my_engine = "my_pkg.backend:MyEngine"
```

Built-in backends always take precedence over plugins of the same name, plugin
discovery is lazy (it never slows cold import), and `list_backends()` returns
the merged builtin + plugin set. Requesting a known-but-uninstalled plugin
backend (e.g. `create_simulation("isaac")` without the plugin) raises a
`ValueError` carrying the exact install hint. See
[`docs/simulation/isaac.md`](docs/simulation/isaac.md) for the Isaac Sim
backend's install, usage, config, and `STRANDS_ISAAC_*` env vars.

## Mesh networking

<p align="center">
  <img src="docs/assets/mesh_network.svg" alt="Strands Robots mesh - robot peers discovering and coordinating over the Zenoh mesh" width="100%">
</p>

Every `Robot()` and `Simulation()` is automatically a peer on a local Zenoh
mesh - no setup. Peers on the same LAN discover each other via multicast
scouting, sharing a single ref-counted `zenoh.Session` per process.

```python
from strands_robots import Robot

a = Robot("so100")              # auto-joins the mesh
b = Robot("so100")              # second peer (another process)
print(a.mesh.peers)             # list[dict] - discovers b
print(a.mesh.peers_by_id[b.peer_id])   # dict[peer_id -> info] for O(1) lookup
info = a.mesh.get_peer(b.peer_id)      # None-safe single lookup

a.mesh.tell(b.peer_id, "pick up the cube")
a.mesh.emergency_stop()         # broadcast E-STOP, audited to disk
```

`tell()` routes to hardware **and** sim peers. Per-call policy kwargs
(`target_pose`, `target_joints`, `world_update`) and constructor extras are
forwarded end-to-end via `policy_config`, so a planner-style policy on a sim
peer sees the goal payload it needs:

```python
a.mesh.tell(
    b.peer_id,
    "reach for the red block",
    policy_provider="curobo",
    target_pose=[0.3, 0.0, 0.4, 1.0, 0.0, 0.0, 0.0],
    robot_name="arm_left",      # disambiguate in multi-robot sims
    duration=10.0,
)
```

Expose the mesh to an agent with the `robot_mesh` tool (`peers`, `status`,
`tell`, `send`, `broadcast`, `stop`, `emergency_stop`, `subscribe`, `watch`,
`inbox`). Disable globally with `STRANDS_MESH=false` or per-robot with
`Robot("so100", mesh=False)`. Install with `uv pip install "strands-robots[mesh]"`.

For frictionless single-machine experiments, set `STRANDS_MESH_LOCAL_DEV=1` -
one env var that runs the mesh without mTLS/ACL on localhost. It defaults the
auth mode to `none` **and** satisfies the insecure-acknowledgement second
factor by itself, so you don't also need `STRANDS_MESH_I_KNOW_THIS_IS_INSECURE=1`.
An explicit `STRANDS_MESH_AUTH_MODE=mtls` still wins. **Never** set
`STRANDS_MESH_LOCAL_DEV` on a shared or production network.

### AWS IoT Core transport (fleets)

For robots across networks, bridge the mesh to AWS IoT Core over MQTT5/mTLS,
with Device Shadow mirroring, S3 camera offload, and account-wide Fleet
Provisioning. Hardened with CA pinning, strict thing-name validation,
deny-by-default IoT policy scoping, and a safety audit log.
Install with `uv pip install "strands-robots[mesh-iot]"`. See the
[Configuration](#configuration) matrix for the `STRANDS_MESH_*` knobs.

## ROS 2 interoperability

strands-robots speaks ROS 2 from four complementary angles - a Strands agent can
observe, command, *be*, and *expose* a ROS 2 system. Full guide:
[ROS 2 Integration](https://strandsagents.com) / `docs/ros2-integration.md`.

![A Strands agent driving a closed-loop square in turtlesim via use_ros](docs/assets/use_ros_agent_square.gif)

*A Strands agent (Claude Opus via Amazon Bedrock) given the `use_ros` tool drives
a real ROS 2 `turtlesim` in a closed-loop square - reading pose, correcting
heading, re-driving - over 43 in-process tool calls. Runnable:
`examples/ros2/use_ros/`.*

| Surface | What it does | Backend | Needs sourced ROS 2 |
|---------|--------------|---------|---------------------|
| **`use_ros`** | List/echo/publish topics, call services on any ROS 2 graph | in-process `rclpy` | yes |
| **`use_rtps`** | Join a graph as a DDS peer and **act as a robot** (publish topics a real stack consumes) | pure `cyclonedds` (pip) | **no** - macOS/CI/Jetson, all distros |
| **`RosBridgedRobot`** | Drive a `cmd_vel`/odom ROS 2 base as a first-class strands `Robot` | `use_ros` | yes |
| **`SimEngine(ros2_bridge=True)`** | Publish a running MuJoCo sim's `joint_states` + camera `image_raw` so rviz/nav2/agents can subscribe | `rclpy` | yes |

```python
# Observe + command a live ROS 2 graph, in plain English:
from strands import Agent
from strands_robots.tools import use_ros
Agent(tools=[use_ros])("list the topics, drive /turtle1 forward, confirm the pose changed")

# Or expose a simulation as a ROS 2 node any tool can subscribe to:
from strands_robots.simulation import Simulation
sim = Simulation(ros2_bridge=True)
sim.create_world(); sim.add_robot("so101")
sim.step(10)   # publishes /so101/joint_states + camera image_raw on the ROS 2 domain
```

`rclpy` ships with a sourced ROS 2 distro (not on PyPI). The `[ros2]` extra adds
only the pip-installable `cyclonedds` binding that `use_rtps` uses - so the
pure-RTPS path needs no ROS install at all. Every surface degrades to a clear,
structured error when its backend is unavailable; the default install never
touches ROS 2.

## Configuration

### Environment variables

| Variable | Description | Default |
|----------|-------------|---------|
| `STRANDS_ROBOT_MODE` | `Robot()` factory mode: `sim` / `real` / `auto` | `sim` |
| `STRANDS_ASSETS_DIR` | Robot model asset cache directory | `~/.strands_robots/assets/` |
| `STRANDS_MEMORY_DIR` | Harness memory store (`harness_memory` tool: task solution traces + global success rules / failure models) | `~/.strands_robots/memory/` |
| `STRANDS_ROBOTS_RENDER_ROOT` | Sandbox directory that `Simulation.render(output_path=...)` may write into | `~/.strands_robots/renders/` |
| `STRANDS_ROBOTS_RENDER_ALLOW_ABS` | Set `1` to allow `render(output_path=...)` to write absolute paths outside the render sandbox | unset |
| `STRANDS_ROBOTS_RENDER_MAX_BYTES` | Max PNG size `render(output_path=...)` will persist | `52428800` (50 MB) |
| `STRANDS_ROBOTS_VIDEO_ROOT` | Opt-in sandbox for video/recording output paths (`run_policy(video=...)`, `start_cameras_recording`). Unset = absolute paths allowed (historic contract); set to confine writes | unset |
| `STRANDS_ROBOTS_VIDEO_ALLOW_ABS` | Set `1` to re-permit absolute paths when `STRANDS_ROBOTS_VIDEO_ROOT` is set | unset |
| `STRANDS_TRUST_REMOTE_CODE` | Set `1` to allow HF `trust_remote_code` for `lerobot_local` | unset |
| `STRANDS_ROBOTS_NO_DYLD_SHIM` | Set `1` to disable the macOS auto-fix that puts Homebrew ffmpeg on the dyld path for torchcodec video streaming (see [Recording & streaming datasets](#recording--streaming-datasets)) | unset |
| `MUJOCO_GL` | MuJoCo GL backend (`egl`, `osmesa`, `glfw`) | auto |
| `STRANDS_ISAAC_HEADLESS` | Isaac Sim backend: run without a GUI (`true`/`1`/`yes` = headless). Overrides `IsaacConfig(headless=...)` | unset (config default `true`) |
| `STRANDS_ISAAC_RTX_PATHTRACING` | Isaac Sim backend: set `true`/`1`/`yes` to enable RTX path-tracing (photorealistic, slow) instead of the default render mode | unset |
| `STRANDS_ISAAC_NUCLEUS_URL` | Isaac Sim backend: override the Omniverse Nucleus asset-server URL | unset (Isaac default) |
| `GROOT_API_TOKEN` | API token for the GR00T inference service | unset |
| `STRANDS_MESH` | Set `false` to disable Zenoh mesh globally | `true` |
| `STRANDS_MESH_LOCAL_DEV` | Set `1` for a one-var localhost preset (auth `none`, no second factor needed) | unset |
| `STRANDS_ROS2_BRIDGE_I_KNOW_THIS_IS_INSECURE` | Second factor to expose a `Robot(ros2_transport="rtps")` inbound `joint_command` surface with no `dds_security_config` (DDS Security). Truthy: `1`/`true`/`yes` | unset |
<details>
<summary><b>Mesh / IoT / GR00T-container env vars (advanced)</b></summary>

| Variable | Description | Default |
|----------|-------------|---------|
| `STRANDS_MESH_AUTH_MODE` | Wire auth: `mtls` or `none` (`none` needs a second factor) | `mtls` |
| `STRANDS_MESH_I_KNOW_THIS_IS_INSECURE` | Second factor required to bring up `AUTH_MODE=none` | unset |
| `STRANDS_MESH_PORT` | TCP port for the local Zenoh router | `7447` |
| `ZENOH_CONNECT` | Comma-separated remote Zenoh endpoints to connect to | unset |
| `ZENOH_LISTEN` | Comma-separated endpoints for the local Zenoh listener | unset |
| `STRANDS_MESH_AUDIT_DIR` | Directory for the safety audit log (`mesh_audit.jsonl`) | `~/.strands_robots/` |
| `STRANDS_MESH_CA_PINS` | Additional SHA-256 CA pins (comma-separated 64-char hex) | unset |
| `STRANDS_MESH_DISABLE_CA_PIN` | Skip CA pin check on download path (break-glass) | `false` |
| `STRANDS_MESH_CAMERA_PRESIGN_TTL` | TTL (s) for S3 presigned camera URLs; capped at 3600 | `60` |
| `STRANDS_MESH_ACL_FILE` | Path to a JSON5 Zenoh ACL file; unset = permissive default. See `examples/mesh/mesh_acl_example.json5` (role-scoped) and `examples/mesh/mesh_acl_strict_per_peer.json5` (per-peer). **⚠️ Required on any WAN/cloud router: mTLS gives identity, not least-privilege — without a topic-level ACL one device cert can read all fleet traffic and command any robot. See [security docs](docs/security.md#production-posture-required-off-trusted-networks).** | unset |
| `STRANDS_MESH_POLICY_HOST_ALLOW` | Comma-separated allowlist of VLA policy-server hosts/CIDRs for inference | loopback only |
| `STRANDS_MESH_HITL_ACTIONS` | `robot_mesh` actions needing a human-in-the-loop interrupt: `all` / `none` / subset of `emergency_stop,broadcast,tell,send,stop,subscribe,watch` | actuation default |
| `STRANDS_MESH_SUBSCRIBE_ALLOW` | Extra Zenoh key-expr patterns the `robot_mesh` `subscribe` action may target, beyond the built-in low-impact set | shared classes only |
| `STRANDS_MESH_OVERRIDE_CODE` | Shared secret for e-stop resume HMAC proof; unset means no remote resume possible | unset |
| `STRANDS_MESH_INPUT_VALUE_ABS` | Absolute value clamp for teleop joint commands (radians) | `12.566` (4pi) |
| `STRANDS_MESH_INPUT_MAX_HZ` | Per-receiver teleop apply-rate ceiling (0 = unlimited). A value no rate check can be built from -- unparsable, or non-finite like `inf`/`nan` -- falls back to the default so the ceiling stays enforced | `100` |
| `STRANDS_MESH_INPUT_SLEW_ABS` | Per-joint speed bound for teleop commands, in frame units per second (widen for degree-valued or normalized actuators; cannot be disabled) | `25.133` (8pi) |
| `STRANDS_MESH_POSE_HZ`, `_IMU_HZ`, `_ODOM_HZ`, `_HEALTH_HZ`, `_LIDAR_SUMMARY_HZ`, `_HAND_HZ`, `_MAP_INFO_HZ` | Per-topic sensor publish rate; `0` (or any non-positive value) switches that topic off. A value the loop cannot pace itself with keeps the built-in rate | per topic: `10`/`10`/`10`/`0.5`/`5`/`50`/`0.2` |
| `STRANDS_MESH_CAMERA_HZ` | Camera publish rate; opt-in because frames are large. Unset, non-positive, or unusable leaves camera publishing off | `0` (off) |
| `STRANDS_MESH_MAX_PEERS` | Peer registry cap; evicts oldest on overflow | `1024` |
| `STRANDS_MESH_RESUME_MAX_FAILS` | Failed resume attempts before cooldown engages | `5` |
| `STRANDS_MESH_RESUME_BACKOFF_S` | Cooldown (seconds) after exceeding resume fail threshold | `30` |
| `STRANDS_MESH_INPUT_AUDIT_EVERY` | Emit `input_stream_applied` audit event every N frames (0 = off) | `100` |
| `STRANDS_ESTOP_DEDUP_TTL_S` | E-stop fan-out Lambda dedup window (seconds) | `30` |
| `STRANDS_MESH_BRIDGE_TOPICS` | Comma-separated topic suffixes the Zenoh<->IoT bridge forwards (exact match). Unset = the safe default set (`presence,health,safety/event,safety/estop,safety/resume,cmd,response,broadcast`). High-volume topics (`state,pose,imu,odom,lidar`) and LAN-only topics (`camera,input,hand`) are deliberately NOT bridged | default set |
| `STRANDS_MESH_BRIDGE_TOPICS_PREFIX` | Comma-separated topic suffixes the bridge matches as a path **prefix** (so `response` matches `response/<turn-id>`). Extend this (not `STRANDS_MESH_BRIDGE_TOPICS`) when adding an RPC-shape topic with a per-turn tail | `response` |
| `STRANDS_GR00T_IMAGE` | Container image the `gr00t_inference` tool runs (must pass the image allowlist; agent cannot choose it) | `gr00t:latest` |
| `STRANDS_GR00T_IMAGE_ALLOW` | Extra image-name patterns (trailing `*` = tag wildcard) added to the built-in allowlist (`gr00t:*`, `nvcr.io/nvidia/isaac-gr00t:*`) | built-in only |
| `STRANDS_GR00T_SERVER_SEED` | Default seed the GR00T determinism wrapper applies at server start and on seedless `reset` calls (used with `gr00t_inference(..., deterministic=True)`; forwarded into the container) | `42` |
| `STRANDS_GR00T_STRICT_DETERMINISTIC` | `1` makes the determinism wrapper additionally enable `torch.use_deterministic_algorithms(True, warn_only=True)` (slower kernels, strictest reproducibility; forwarded into the container) | `0` |

</details>

<details>
<summary><b>Isaac Sim backend env vars (<code>strands-robots[sim-isaac]</code>)</b></summary>

These are read by the built-in, in-tree Isaac Sim backend
(`pip install 'strands-robots[sim-isaac]'`) when it builds its
`IsaacConfig`; an explicit `create_simulation("isaac", ...)` kwarg always wins.
See [`docs/simulation/isaac.md`](docs/simulation/isaac.md).

| Variable | Description | Default |
|----------|-------------|---------|
| `STRANDS_ISAAC_NUCLEUS_URL` | Override the Omniverse Nucleus server URL (when `nucleus_url` is not passed) | unset (Isaac defaults) |
| `STRANDS_ISAAC_HEADLESS` | Truthy (`1`/`true`/`yes`) forces headless; falsy forces a window | unset (uses `headless` kwarg) |
| `STRANDS_ISAAC_RTX_PATHTRACING` | Truthy forces `render_mode="rtx_pathtracing"` | unset |

</details>

<details>
<summary><b>Benchmark / diagnostic env vars (LIBERO, GR00T bisection)</b></summary>

| Variable | Description | Default |
|----------|-------------|---------|
| `STRANDS_LIBERO_ACTION_LOG` / `_MAX` | Per-step OSC controller diagnostics | unset / `50` |
| `STRANDS_LIBERO_STATE_LOG` / `_MAX` | Per-step state values fed to GR00T | unset / `50` |
| `STRANDS_GROOT_WIRE_LOG` / `_MAX_CALLS` | Dump pre/post inference payloads to verify LOCAL vs SERVICE parity | unset / `10` |

</details>

### Asset cache

```
~/.strands_robots/
└── assets/           # auto-downloaded MJCF + meshes
    ├── trs_so_arm100/
    ├── franka_emika_panda/
    └── ...
```

Clear with `rm -rf ~/.strands_robots/assets/`; relocate with
`export STRANDS_ASSETS_DIR=/path/to/dir`.

## Benchmarks

`strands-robots` ships a [LIBERO](https://github.com/Lifelong-Robot-Learning/LIBERO)
benchmark integration on the MuJoCo backend - byte-equivalent to upstream
LIBERO at the model level, reaching `success_rate >= 0.92` on libero-10/SCENE5.
Register declarative benchmarks from file and evaluate policies via the
`list_benchmarks`, `register_benchmark_from_file`, and `evaluate_benchmark`
simulation actions. Install with `uv pip install "strands-robots[benchmark-libero]"`.

## Project structure

```
strands_robots/
├── __init__.py            # Lazy-loaded public API (Robot, Simulation, policies)
├── robot.py               # Robot() factory (sim/real/auto dispatch)
├── hardware_robot.py      # HardwareRobot - async LeRobot control
├── policies/
│   ├── base.py            # Policy ABC
│   ├── factory.py         # create_policy() + runtime registration
│   ├── mock.py            # MockPolicy (non-VLA reference)
│   ├── groot/             # NVIDIA GR00T (ZMQ/HTTP client + data configs)
│   └── lerobot_local/     # Direct HuggingFace inference (RTC, processors)
├── registry/              # robots.json (70+) + policies.json + loaders
├── simulation/
│   ├── base.py            # SimEngine ABC
│   ├── factory.py         # create_simulation() + backend registry
│   ├── models.py          # SimWorld / SimRobot / SimObject / SimCamera
│   └── mujoco/            # MuJoCo backend (67-action AgentTool)
├── rendering/             # Hybrid rendering: CameraParams, backgrounds (panorama/3DGS),
│                          #   HybridCompositor, encode_clip / mjpeg_frames
├── mesh/                  # Zenoh mesh: core, sensors, input, audit, transport, iot
├── benchmarks/libero/     # LIBERO suite + BDDL parser + adapter
└── tools/                 # gr00t_inference, lerobot_*, pose, serial, robot_mesh
```

## Development

```bash
uv pip install -e ".[all,dev]"

hatch run test          # unit tests
hatch run test-integ    # integration tests (GPU + model weights)
hatch run lint          # ruff check + format --check + mypy
hatch run format        # ruff check --fix + ruff format
```

Python 3.12+ required. See [AGENTS.md](AGENTS.md) for conventions and the
accumulated code-review learnings.

## Security

Found a vulnerability? **Do not** open a public issue. Follow the disclosure
process in [SECURITY.md](SECURITY.md) (AWS VDP / HackerOne).

Note the `trust_remote_code` gate on `lerobot_local` (see
[Policy providers](#policy-providers)) and the mesh CA-pinning / thing-name
validation controls in the [Configuration](#configuration) matrix.

## Contributing

Issues and PRs welcome. Track work on the
[Strands Labs - Robots project board](https://github.com/orgs/strands-labs/projects/2);
it is the source of truth for roadmap and follow-ups.

- [GitHub Issues](https://github.com/strands-labs/robots/issues)
- [Pull Requests](https://github.com/strands-labs/robots/pulls)

## License

Apache-2.0 - see [LICENSE](LICENSE).

## Links

<div align="center">
  <a href="https://github.com/strands-labs/robots">GitHub</a>
  ◆ <a href="https://pypi.org/project/strands-robots/">PyPI</a>
  ◆ <a href="https://github.com/google-deepmind/mujoco">MuJoCo</a>
  ◆ <a href="https://github.com/NVIDIA/Isaac-GR00T">NVIDIA GR00T</a>
  ◆ <a href="https://github.com/huggingface/lerobot">LeRobot</a>
  ◆ <a href="https://strandsagents.com/">Strands Docs</a>
</div>
