OopsieVerse provides a damage-aware, simulator-agnostic framework and benchmark for evaluating and learning safer policies
- Clone and enter the repository:
- Linux / macOS:
git clone https://github.com/UT-Austin-RobIn/oopsieverse.git && cd oopsieverse - Windows:
git clone https://github.com/UT-Austin-RobIn/oopsieverse.git; cd oopsieverse
- Linux / macOS:
- Run the installer with
python install.py --new_envand the simulators you need--robocasaor--behavior1k conda activate oopsieversepip install -e .- test OG installation by:
python -m omnigibson.examples.robots.all_robots_visualizer - test Robocasa installation by:
python -m robocasa.demos.demo_kitchen_scenes
- teleop (b1k)
python scripts/teleop_b1k.py --task_name shelve_item --live_feedback --save_video- playback (b1k)
python scripts/playback_b1k.py --task_name shelve_item --collect_hdf5_path demos/behavior1k/teleop_data/shelve_item.hdf5 --playback_hdf5_path demos/behavior1k/playback_data/shelve_item.hdf5 --playback --visualize --compute_metricsThe NVIDIA driver on your system is too old (found version 12080). Please update your GPU driver by downloading and installing a new version from the URL: http://www.nvidia.com/Download/index.aspx Alternatively, go to: https://pytorch.org to install a PyTorch version that has been compiled with your version of the CUDA driver.Uninstall torch and re-install a version compatible with your CUDA. We do:
pip uninstall torch
pip install torch==2.9.1
For using spacemouse:
Add this for spacemouse compact to /etc/udev/rules.d/99-spacemouse.rules
KERNEL=="hidraw*", ATTRS{idVendor}=="256f", ATTRS{idProduct}=="c635", MODE="0666", GROUP="plugdev"
SUBSYSTEM=="usb", ATTRS{idVendor}=="256f", ATTRS{idProduct}=="c635", MODE="0666", GROUP="plugdev"
then run sudo udevadm control --reload-rules
- Works with pyspacemouse==1.1.4 (did not work with pyspacemouse==2.0.0)
- install telemoma for b1k
- force use mediapipe==0.10.21 (newer version gives an error related to "solutions")