Skip to main content

Teleop Project (x86_64 PC)

Use venv/teleop for Pico / XR body tracking and realtime retarget inspection on a laptop or desktop. Record and replay motion clips from the root project with the any4hdmi-backed scripts.

Setup​

uv --project venv/teleop sync

Install XRoboToolkit PC Service​

Download the .deb from:

https://github.com/XR-Robotics/XRoboToolkit-PC-Service/releases

sudo apt install -y ./XRoboToolkit_PC_Service_*.deb

Start XRoboToolkit / XRobot from the desktop or app launcher after installation.

Clone the Pico SDK repos​

mkdir -p external
git clone https://github.com/YanjieZe/XRoboToolkit-PC-Service-Pybind.git \
  external/XRoboToolkit-PC-Service-Pybind
git clone https://github.com/XR-Robotics/XRoboToolkit-PC-Service.git \
  external/XRoboToolkit-PC-Service

Build and install xrobotoolkit_sdk​

bash scripts/setup/setup_xrobot_pybind.sh

The script expects the two repos above to exist under external/.

Enable Pico streaming​

  1. Wear the leg trackers.
  2. Finish whole-body tracking calibration.
  3. Open XRoboToolkit, connect to this machine, and enable whole-body streaming.

Verify Installation​

Start the live retarget publisher:

uv --project venv/teleop run sim2real/teleop/pico_retarget_pub.py

Open the mjviser URL printed by the publisher. If it updates with live G1 retargeted motion, the teleop stack is ready.

To record and replay a qpos clip after the live publisher is running, switch to the root project environment:

uv run scripts/record_motion.py
uv run scripts/view_motion.py --motion g1_motion_YYYYMMDD_HHMMSS/motions/motion.npz

Next Steps​