Skip to main content

Environment Setup

This guide provides high-level instructions for setting up your development environment for Physical AI and Humanoid Robotics development. For detailed, step-by-step instructions, always refer to the official documentation of each tool.

Operating System​

Recommended: Ubuntu 22.04 LTS (Jammy Jellyfish)

Ubuntu 22.04 LTS is the standard platform for ROS 2 Humble and most robotics tools. You have several options:

Option 1: Native Ubuntu Installation​

  • Install Ubuntu 22.04 LTS as your primary OS or dual-boot
  • Best performance and compatibility

Option 2: Virtual Machine​

  • Use VirtualBox, VMware, or Hyper-V
  • Allocate at least 4GB RAM and 50GB disk space
  • Enable hardware acceleration if available

Option 3: WSL 2 (Windows)​

  • Windows Subsystem for Linux 2 on Windows 10/11
  • Good for development, but limited for GPU acceleration
  • May require additional setup for GUI applications

Core Software Stack​

1. ROS 2 (Humble Hawksbill or Iron Irwini)​

ROS 2 is the robotic middleware that forms the "nervous system" of your robot.

Installation:

  • Follow the official ROS 2 installation guide: https://docs.ros.org/en/humble/Installation.html
  • Choose either Humble (LTS) or Iron (latest) based on your needs
  • Ensure you source the setup script in your .bashrc: 
    echo "source /opt/ros/humble/setup.bash" >> ~/.bashrc

Verification:

ros2 --help

2. Gazebo (Classic or Garden)​

Gazebo provides physics simulation for robots.

For ROS 2 Humble:

  • Gazebo Classic (recommended for compatibility)
  • Install via: sudo apt install ros-humble-gazebo-ros-pkgs

For ROS 2 Iron:

Verification:

gazebo --version

3. Unity (Optional)​

Unity is used for visualization and human-robot interaction scenes.

Installation:

  • Download Unity Hub: https://unity.com/download
  • Install Unity Editor (2021.3 LTS or newer)
  • Install ROS-TCP-Connector package for ROS 2 integration

Note: Unity is optional for basic development but recommended for advanced visualization.

4. NVIDIA Isaac Sim & Isaac ROS​

NVIDIA Isaac provides photorealistic simulation and AI perception pipelines.

Prerequisites:

  • NVIDIA GPU with CUDA support (RTX 2060 or better recommended)
  • NVIDIA drivers (version 470+)
  • Docker (for Isaac ROS)

Isaac Sim Installation:

Isaac ROS Installation:

Verification:

docker run --rm nvcr.io/nvidia/isaac/isaac-ros:latest

Development Paths​

Path 1: Local Workstation Development​

Hardware Requirements:

  • CPU: 6+ cores (Intel i7/AMD Ryzen 7 or better)
  • RAM: 16GB minimum, 32GB recommended
  • GPU: NVIDIA RTX 3060 or better (for Isaac Sim)
  • Storage: 100GB+ free space (SSD recommended)

Best for:

  • Learning and development
  • Simulation-heavy work
  • Cost-effective for individuals

Path 2: Cloud + Edge Deployment​

Cloud Workstation (e.g., AWS g5/g6e instances):

  • GPU-enabled cloud instances for training and simulation
  • Cost: ~$1-3/hour depending on instance type
  • Best for: Teams, cost-sharing, or when local hardware is limited

Edge Device (NVIDIA Jetson):

  • Jetson Orin Nano/NX for deployment
  • RealSense cameras, IMU, microphone arrays
  • Best for: Real robot deployment and sim-to-real transfer

Pattern: Train-in-Cloud, Deploy-to-Jetson

  • Develop and train in cloud workstations
  • Deploy optimized models to Jetson edge devices
  • Avoids the "latency trap" of cloud-based robot control

The Latency Trap​

Why controlling real robots from the cloud is dangerous:

  • Network latency (50-200ms+) makes real-time control impossible
  • Safety-critical systems require sub-10ms control loops
  • Connection drops can cause robot failures or safety hazards

Solution:

  • Train models in the cloud (fast iteration, powerful GPUs)
  • Deploy inference to edge devices (low latency, reliable)
  • Use cloud for monitoring and data collection, not control

Quick Setup Checklist​

  • Install Ubuntu 22.04 LTS
  • Install ROS 2 Humble or Iron
  • Install Gazebo
  • (Optional) Install Unity Hub and Editor
  • Install NVIDIA drivers (if using GPU)
  • Install Docker (for Isaac ROS)
  • Install Isaac Sim via Omniverse Launcher
  • Verify all installations with test commands
  • Set up ROS 2 workspace: mkdir -p ~/ros2_ws/src

Getting Help​

If you encounter issues:

  1. Check official documentation first - Most tools have excellent docs
  2. ROS 2 Discourse: https://discourse.ros.org/
  3. Gazebo Forums: https://community.gazebosim.org/
  4. NVIDIA Isaac Forums: https://forums.developer.nvidia.com/c/omniverse/isaac-sim/

Next Steps​

Once your environment is set up: