dora-openarm-mujoco

MuJoCo simulation node for the OpenArm bimanual robot, designed to run inside a dora-rs dataflow.

It replaces the physical follower arms and cameras: it accepts joint-position commands, publishes canonical arm position/state snapshots on request, and renders JPEG camera frames.

Installation

uv sync

Quick start

A self-contained dummy dataflow is included for testing without real hardware. It wires a dummy leader node to the MuJoCo sim and records to disk.

uv run dora build dataflow-dummy.yaml --uv
uv run dora run dataflow-dummy.yaml

Dataflow configuration

Minimal (headless and position forwarding only no cameras)

- id: openarm-mujoco
  build: pip install -e .
  path: dora-openarm-mujoco
  inputs:
    move_position_right: leader/follower_position_right
    move_position_left:  leader/follower_position_left
    request_position: dora/timer/millis/4
  outputs:
    - status
    - position_right
    - position_left

Full (interactive viewer + all cameras, with contacts, can be used for vr teleoperation)

- id: openarm-mujoco
  build: pip install -e .
  path: dora-openarm-mujoco
  args: "--viewer --render --enable-collision --ctrl --keyframe home"
  inputs:
    move_position_right: leader/follower_position_right
    move_position_left:  leader/follower_position_left
    request_state: dora/timer/millis/4
  outputs:
    - status
    - state_right
    - state_left
    - camera_wrist_right
    - camera_wrist_left
    - camera_head_left
    - camera_head_right
    - camera_ceiling

Inputs

Outputs

Camera outputs carry metadata={"encoding": "jpeg"}.

State diagnostics are simulation placeholders, not hardware health measurements. motor_status follows qpos order: joints 1-7 then finger_joint1, with ENABLED for existing joints and SILENT for missing joints. bus.carrier is always true (bool); the cumulative counters bus_off, error_passive, error_warning, ack_error, tx_overflow, rx_overflow, and net_down are always zero (int64). The simulator does not model CAN communication, motor faults, or temperature.

Arm snapshots are published only on request_position or request_state, including before the first position command. Connect the desired request input to a timer independent of control outputs. request_position publishes only position_*; request_state publishes only state_*, which includes qpos, velocity, and actuator force. Startup publishes ready; position commands only update the simulation target. Each request samples both arms under one lock. After sampling both arms, the node captures one time.time_ns() timestamp before releasing the lock and uses it as observation_timestamp for every output of that request. This is integer Unix wall-clock nanoseconds, not the request time or MuJoCo simulation time (data.time). Request metadata is copied without modification except for timestamp, which is removed from the copy so Dora supplies each output message timestamp.

To migrate an older dataflow, rename command inputs from position_* to move_position_*, replace MuJoCo's arm_right_observation and arm_left_observation outputs with position_* or state_*, and add a request_position or request_state input. The old output ports have been removed. Consumers receive Arrow structs rather than flat arrays. The dummy example connects recorder's arm_right_observation / arm_left_observation inputs to state_right / state_left; these recorder input names remain unchanged.

Arguments

Pass these via the args: field in the dataflow YAML, or directly on the command line.

Environment variables

Every argument's default can also be set via an environment variable named DORA_OPENARM_MUJOCO_ + the upper-cased argument name:

Boolean flags accept 1/0, true/false, yes/no and on/off. DORA_OPENARM_MUJOCO_VIEWER additionally accepts an FPS number (e.g. 60); true enables the viewer at the default 30 Hz.

Explicit CLI arguments override the environment. Boolean flags also gain a --no-* form (e.g. --no-render, --no-viewer) to turn an environment-enabled option back off. Set the variables via the env: field of the node in the dataflow YAML:

- id: openarm-mujoco
  path: dora-openarm-mujoco
  env:
    DORA_OPENARM_MUJOCO_VIEWER: "60"
    DORA_OPENARM_MUJOCO_RENDER: "true"

License

Licensed under the Apache License 2.0. See LICENSE for details.

Copyright 2026 Enactic, Inc.

Code of Conduct

All participation in the OpenArm project is governed by our Code of Conduct.