Skip to main content

Installation

Teleopit supports multiple installation profiles depending on your use case.

Prerequisites​

  • Python 3.10+
  • Conda (recommended)
conda create -n teleopit python=3.10
conda activate teleopit

Install Profiles​

Inference Only (sim2sim)​

pip install -e .

This is sufficient for offline BVH playback and MuJoCo simulation.

Training​

pip install -e '.[train]'

Adds rsl-rl-lib, mjlab, wandb, and training dependencies.

Sim2Real (Hardware Deployment)​

pip install -e '.[sim2real]'

Adds opencv-python and g1_bridge_sdk. You also need to initialize submodules and build the C++ bridge:

git submodule update --init --recursive
bash scripts/setup/setup_g1_bridge.sh

See G1 Bridge SDK for details.

Pico 4 VR​

pip install -e '.[pico4]'

Teleopit uses the in-process pico_bridge.PicoBridge receiver for Pico tracking. For Teleopit 0.3.0, use pico-bridge 0.2.0. Do not upgrade the receiver to pico-bridge 0.2.1, because 0.2.1 changes interface semantics that Teleopit 0.3.0 does not target. The receiver can run on a workstation PC or the robot onboard computer. See Pico Sim2Sim and Pico Sim2Real for the full setup guides.

Verify Installation​

python -c "import teleopit; print('teleopit OK')"
python -c "import train_mimic.tasks; print('training OK')"  # if training installed

Next Steps​