motion-stack-2

Install

pixi install
pixi run router  # start zenoh router (required)

Examples

All examples use the Moonbot Zero (4 legs, 3 joints each, 12 joints total). Communication uses pyzeros -- real ROS 2 topics over zenoh, fully interoperable with rclpy, rviz, rosbag, etc. No ROS installation required.

Quick start: run from the terminal

No Python code needed. Compile the URDF, then launch:

# compile the URDF
pixi run python -c "from ms_moonbot_zero import load_moonbot_zero_urdf; \
  open('/tmp/mz.urdf','w').write(load_moonbot_zero_urdf())"

# run the robot
pixi run python -m ms_pyzeros_bridge.lvl1_exec \
  --ms-lvl1-json '{"urdf": "/tmp/mz.urdf", "namespace": "ms"}'

# (other terminal) visualize in rerun
pixi run python -m ms_pyzeros_bridge.rerun_viz --urdf /tmp/mz.urdf

Works with any URDF. The urdf field accepts a file path or string.

Note

Mesh paths inside the URDF must be absolute URIs (file:///...), not ROS package URIs (package://...), since there is no ament_index at runtime.

Single-process robot (robot_demo.py)

Low-level approach: you wire up JointCore, transport, and simulation yourself. Use this when you need custom hardware drivers, processing hooks, or a different transport.

pixi run python -m ms_moonbot_zero.robot_demo
pixi run python -m ms_moonbot_zero.robot_demo --rerun

Note

Shortcut: pixi run mz_robot / pixi run mz_robot --rerun

Tip

The robot starts idle. Send commands with the API client below, or with ROS 2 CLI:

ros2 topic list
ros2 topic echo /joint_read
ros2 topic pub /joint_set sensor_msgs/msg/JointState \
  "{name: ['joint1_1', 'joint1_2'], position: [0.5, -0.3]}"

Multi-process launcher (mz_launcher.py)

Same idea as the terminal quick start, but automated with Popen: compile the URDF in Python, build Lvl1Param per leg, and spawn one lvl1_exec process each. Use this when the defaults work and you just need to configure and launch.

pixi run python -m ms_moonbot_zero.mz_launcher
pixi run python -m ms_moonbot_zero.mz_launcher --viz

Note

Shortcut: pixi run mz_launcher / pixi run mz_launcher --viz

Each leg publishes on its own namespace (leg1/joint_read, leg2/joint_read, etc.).

API client (api_demo.py)

Connects to a running robot and moves all joints via AsyncJointSyncer.

# terminal 1: start the robot
pixi run python -m ms_moonbot_zero.robot_demo
# or: pixi run python -m ms_moonbot_zero.mz_launcher

# terminal 2: start the API client
pixi run python -m ms_moonbot_zero.api_demo

Note

Shortcuts: pixi run mz_robot, pixi run mz_launcher, pixi run mz_api

Works with both single-process and multi-process mode.