cookbook

Cookbook

OSMO cookbook is a comprehensive suite of AI workflows to perform a wide variety of robotics tasks.

Tutorials

Collection of workflows from the tutorial documentation

How-To Guides

Workflow Name	Link
Isaac Sim: Generating Synthetic Data	README
Isaac Lab: Training Robot Policy with Reinforcement Learning	README
ROS2: Multi-Node Communication	README
TorchRun: Training on a Single Node	README
TorchRun: Training on Multiple Nodes	README
TorchRun: Training with Rescheduling	README
Hardware-in-the-Loop: Deploying Policy on Jetson	README

Physical AI End-to-End Pipeline: Nut Pouring

End-to-end pipeline for fine-tuning the GR00T VLA model using the Nut Pouring task: from teleop data through MimicGen, Cosmos Transfer, and LeRobot to a trained policy. Check out the README for more information about this workflow series.

Remote Development

Workflow Name	Link
JupyterLab: Host a Notebook	README
Filebrowser: Launching a Workspace	README
VSCode: Spin up a Remote Server	README
Github: Cloning a Private Repository	README
Ray: Create a Cluster	README
Weights & Biases: Neural Network Training	README
Isaac Sim: Livestreaming Client	README

Additional Applications

Workflow Name	Link
Isaac Lab: Multi-GPU Reinforcement Learning Training	README
Isaac Lab: Multi-Node Reinforcement Learning Training	README
Isaac Groot: Interactive Notebook for Inference and Fine-tuning	README
Isaac Groot: Finetuning a Model	README
Isaac Groot: Imitation Learning using Groot Mimic	README
ROS2: Running Simulation with Foxglove Visualization	README
Gazebo: Generating Synthetic Data	README
Cosmos Predict: Video2World Generation	README
Cosmos: Transfer2.5 with Isaac Sim Integration	README
Cosmos: Video Reasoning and Analysis	README
NIMs: Using NVIDIA NIMs in a workflow	README
TorchRun: Elastic Training on Multiple Nodes	README
DeepSpeed: Training on Multiple Nodes	README