Reinforcement Learning-Controlled Robotics Simulation for Spot and Spot Arm based on NVIDIA Isaac Sim and Isaac Lab

This project provides example applications, pre-trained policies, and environment configurations for research and prototyping of autonomous quadruped and manipulator behaviors in complex environments such as warehouses.

The system leverages the power of Isaac Sim physics simulation and bridges it with ROS 2 for real-time testing and robotic development workflows.

• ✅ Isaac Sim simulation environments for Spot and Spot Arm
• ✅ RL policy controllers for locomotion
• ✅ Example applications including warehouse navigation
• ✅ ROS 2 bridge support for interfacing with external systems
• ✅ Modular and extensible structure for adding new robot models and policies

• NVIDIA Isaac Sim
• ROS 2 Humble
• rmw_zenoh
• GPU: NVIDIA RTX 40xx or better

1. Clone the repository:

git clone https://github.com/mschweig/IsaacRobotics.git

2. Install Isaac Sim and required extensions (refer to Isaac Sim documentation).

3. (Optional) Install ROS 2 Humble and enable isaacsim.ros2.bridge extension in Isaac Sim.

cd /workspaces/IsaacSim
./python.sh /workspaces/IsaacRobotics/applications/spot_warehouse.py

Control the robot via keyboard:

The applications/spot_policy.py contains example implementations of PolicyControllers for Spot and Spot Arm. You can adapt these for your custom simulation experiments.

This project is licensed under the Apache License 2.0. See LICENSE for details.

NVIDIA proprietary code (e.g., RL policies) remains under NVIDIA's licensing terms.