markdowndocs

View project on GitHub

up

Nvidia Robotics / Simulation Suite

  • Note it properly has double a, not a double s

Issac Sim Tutorials

  • https://docs.omniverse.nvidia.com/isaacsim/latest/introductory_tutorials/index.html

Ubuntu

https://docs.omniverse.nvidia.com/launcher/latest/it-managed-launcher/install_guide_linux.html

Code locations

  • Isaac Sim Extensions (all of them): C:\Users\mike\AppData\Local\ov\pkg\isaac_sim-2023.1.1\exts>
  • Isaac Sim Core: cd \Users\mike\AppData\Local\ov\pkg\isaac_sim-2023.1.1\exts\omni.isaac.core\omni\isaac\core\
  • Isaac Sim Motion Generation: cd \Users\mike\AppData\Local\ov\pkg\isaac_sim-2023.1.1\exts\omni.isaac.motion_generation\omni\isaac\motion_generation\
  • Isaac Sim Examples (from the menu): cd \Users\mike\AppData\Local\ov\pkg\isaac_sim-2023.1.1\exts\omni.isaac.examples\omni\isaac\examples\
  • Isaac Sim franka: cd \Users\mike\AppData\Local\ov\pkg\isaac_sim-2023.1.1\exts\omni.isaac.franka\omni\isaac\franka\
  • Isaac Sim ur10: cd \Users\mike\AppData\Local\ov\pkg\isaac_sim-2023.1.1\exts\omni.isaac.universal_robotics\omni\isaac\universal_robotics\
  • Search all those: findstr /a:e /i /s palletizing *.py
  • Better search: grep -i -r "Isaac Examples" . --include=*.py --color=always
  • Palette example: omni.isaac.examples\omni\isaac\examples\ur10_palletizing\ur10_palletizing_extension.py
  • Isaac Sim Template (for testing?): /omni.isaac.ui_template/omni/isaac/ui_template/extension.py

Initializtion Complexity

  • ALl the Isaac Sim example use the Isaac class world, code can be found in Isaac Sim Core

Building a Robot from scratch

  • An old video, but comprehensive ()

Articulations

  • https://docs.omniverse.nvidia.com/py/isaacsim/source/extensions/omni.isaac.core/docs/index.html

Cortex

  • Most of the cool examples use cortex.

UR10 Palletizing

  • One of the coolest demos
  • Code for this is scattered: -====

Surface Grippers

  • Finally got things working with the Jaka Surface Gripper
  • They require an a prim where the end effector is with a rigid body.
    • It is not necessarily part of the articulation.
    • An offset can be specified for the actual contact point.
    • There are a couple ways to specify the offset depending on what level you

Physics Tutorial Notes

  • How to turn on the Physics Toolbar? (Tools/Physics Autoring) or (Window/Simulation/Physics Authoring Toolbar)
  • How to make automatic static mesh creation? (gear)
  • How to get the Simulation Settings diaglog box? (I don’t now)
  • How to group objects? (Create an XForm and create them under it)
  • How to make a group of objects selectable as one object? (chnge kind to “Component” or anything else)
  • How to toggle that selection mode (use the All Model Kinds icon (top left ))

Venv on windows 11 for Python 3.10

  • Had Python 3.13 installed on Windows 11
  • Created the directory for the home of the vevn (d:\ov\isaac_lerobot)
  • cd \ov\isaac_lerobot
  • pip install virtualenv
  • pip install virtualenvwrapper-win
  • python -m virtualenv venv --python=3.10
  • venv\activate
  • Prompt changed to (venv) D:\ov\isaac_lerobot>
  • started python and it was indeed python 3.10.11

Isaac Lab 2.0

  • Needs Isaac Sim 4.5
  • Link to installation docs: (https://isaac-sim.github.io/IsaacLab/main/source/setup/installation/pip_installation.html)
  • Creation of virtual env is wrong instead of:
    • You need this beforehand: sudo apt install python3.10-venv
    • python3.10 -m venv env_isaaclab
    • source env_isaaclab/bin/activate
    • Check CUDA version: nvcc --version - I had version 11.5
      nvcc: NVIDIA (R) Cuda compiler driver
      Copyright (c) 2005-2021 NVIDIA Corporation
      Built on Thu_Nov_18_09:45:30_PST_2021
      Cuda compilation tools, release 11.5, V11.5.119
      Build cuda_11.5.r11.5/compiler.30672275_0
      
    • Install CUDA with correct version
    • Here CUDA 11 or 12, the minor version does not seem to matter)
      • pip install torch==2.5.1 --index-url https://download.pytorch.org/whl/cu118
      • pip install torch==2.5.1 --index-url https://download.pytorch.org/whl/cu124
      • takes like 5 min
      • You could test with
        • python
        • >>> import torch
        • >>> torch.cuda.is_available()
        • True
        • Note that I needed to do a pip install numpy to get this to work
        • you might want to skip this because installing isaac sim coming next will want a particular version of numpy
    • Upgrade pip
      • pip install --upgrade pip
    • Install IsaacSim
      • pip install isaacsim[all,extscache]==4.5.0 --extra-index-url https://pypi.nvidia.com
      • It should run now with isaacsim - but takes about 10 minutes the first time as it is “pulling” additional extensions
      • To see where it installed to a pip show isaacsim

load PushT scene into Isaac sim

c:
cd \temp\testisaacsim
venv\Scripts\activate
Open d:/ov/isaac_lerobot/pusht-scene.usd
Force additiona collider generation (no)

IsaacSim and LeRobot

  • YouTube example: https://www.youtube.com/watch?v=eO5wMzw9LeQ
  • Project download (zip file): https://drive.google.com/drive/folders/1eyh1GzMGnrjCnwW0_CWqEyVtDDd-0Sfe
  • UR Moveit configs for all UR robots

Install lerobot

git clone https://github.com/huggingface/lerobot
pip install -e .

Installation of ROS2 part

 2006  sudo apt update && sudo apt install curl -y
 2009  sudo echo "deb [arch=$(dpkg --print-architecture) signed-by=/usr/share/keyrings/ros-archive-keyring.gpg] http://packages.ros.org/ros2/ubuntu $(. /etc/os-release && echo $UBUNTU_CODENAME) main" | sudo tee /etc/apt/sources.list.d/ros2.list > /dev/null
 2010  sudo apt install ros-humble-moveit-common
 2011  sudo apt install ros-humble-control-msgs
 2012  sudo apt install ros-humble-control-toolbox
 2016  sudo apt install ros-humble-moveit-core
 2017  sudo apt install ros-humble-moveit-servo
 2018  sudo apt install ros-humble-ros-testing
 2019  colcon build

This should bring up Rviz2 with a model of the UR5 arm mounted on a pedestal. Note it only works with a joystick

source ur5_simulation/install/setup.bash
sudo apt install ros-humble-controller-manager
sudo apt install ros-humble-moveit-ros-visualization
ros2 launch ur5_moveit_config arm_joy_control.launch.py

To train a model

todo

To run a trained model

- find `ur_simulation/lerobot_related/examples/2_evaluate_policy.py
- make sure that your policy path points to the directory where your model.safetensors is
source ur5_simulation/install/setup.bash
ros2 launch ur5_moveit_config arm_diffusion_control.launch.py
start your isaac sim simulator

python 2_evaluate_policy.py