Installation ============ .. contents:: :local: :depth: 2 Clone the repository -------------------- .. code-block:: console $ git clone https://github.com/MRRP-lab/arm-demos.git $ cd arm-demos .. note:: If you are on Windows, first install Ubuntu 22.04 with WSL and then clone the repository onto your WSL filesystem. .. code-block:: doscon > wsl --install Ubuntu-22.04 After installation is complete and you are logged in to the WSL system, run the clone commands from above. Install dependencies --------------------- Use the instructions from https://pixi.sh/latest/ to install ``pixi``. Once ``pixi`` has been installed, close the terminal session and start it again, which will ensure ``pixi`` is on the PATH. Install the required dependencies into an isolated development environment: .. code-block:: console $ pixi install You will get a warning. This is expected: .. code-block:: console WARN Could not find activation scripts: ./install/setup.sh This is ``pixi`` trying to include our packages in the shell environment. This will not work until we build our source code. This is done in the next step. Build the source code --------------------- .. code-block:: console $ pixi run build This will build and create an `editable install `__ of the packages from the ``src`` folder. This command should be re-run if metadata is changed or a new file is added to the ``src`` folder. Updates to existing files should not need a rebuild to take effect. .. note:: This will build the project's source code into the ``install`` folder. It will also create the missing ``install/setup.sh`` script causing the previous warning. The ``setup.sh`` script is used to make the packages in the ``install`` folder available for use in the current shell. Pixi will automatically handle calling this script. Download the MediaPipe Model ---------------------------- This project uses hand tracking that depends on a model that must be downloaded separately. A task is available to perform the download: .. code-block:: console $ pixi run download-mediapipe-model Alternatively, you can `download the model yourself `__. The file should be named ``hand_landmarker.task`` in the project root directory. Next steps ---------- See :doc:`Camera Connection `.