Skip to main content

Physical AI & Humanoid Robotics Textbook

From Simulation to Reality - A Spec-Driven Approach​

Welcome to the Physical AI & Humanoid Robotics Textbook, a comprehensive guide to building intelligent robotic systems using ROS 2, modern simulation platforms, and AI integration.

🎯 What Makes This Textbook Unique?​

This isn't just another robotics textbook. It was created using Spec-Kit Plus and Claude Code to demonstrate AI-powered, spec-driven development. Every chapter follows a systematic workflow:

/sp.specify → /sp.plan → /sp.tasks → /sp.implement

All decisions are documented in Architecture Decision Records (ADRs), all development steps are captured in Prompt History Records (PHRs), and all content is validated against a Constitution that ensures quality and consistency.

→ Learn about the Spec-Driven Workflow

📚 Textbook Contents​

Part I: Foundations​

  1. Introduction to Physical AI ✅

    • Embodied AI principles, course toolchain, hardware platforms

  2. ROS 2 Fundamentals 🟡

    • Pub/sub, services, actions, transforms (foundation complete, labs in progress)

  3. Simulation Environments 🚧

    • Gazebo, Unity, Isaac Sim (planned, not yet implemented)

Part II: Core Skills​

  1. Perception Systems 🚧

    • Cameras, LiDAR, sensor fusion (spec complete)

  2. SLAM & Navigation 🚧

    • Mapping, localization, Nav2 (spec complete)

  3. Manipulation & Grasping 🚧

    • MoveIt 2, grasp planning (spec complete)

Part III: Advanced Topics​

  1. Deep Learning for Perception 🚧

    • Vision models, object detection (spec complete)

  2. Control Systems 🚧

    • PID, MPC, whole-body control (spec complete)

  3. Advanced Topics 🚧

    • Multi-robot systems, cloud robotics (spec complete)

  4. Physical Deployment 🚧

    • Jetson setup, sim-to-real transfer (spec complete)

Legend: ✅ Complete | 🟡 Partial | 🚧 Planned

🛠️ Toolchain​

This textbook uses industry-standard tools locked for academic year 2025-2026:

  • ROS 2: Humble Hawksbill (LTS until 2027)
  • Simulation: Gazebo Classic 11+, Gazebo Harmonic, Unity 2022.3 LTS, Isaac Sim 2023.1.1+
  • Compute: Ubuntu 22.04, NVIDIA Jetson Orin, RTX 3060+ recommended
  • AI/ML: Whisper v3, GPT-4 Turbo, YOLO v8, RT-2 VLA models
  • Hardware: Unitree Go2/G1, UBTech Walker Mini (optional)

→ Full toolchain details in Constitution

🎓 Learning Outcomes​

Upon completing this textbook, you will be able to:

  1. Embodied Intelligence: Explain how physical constraints shape AI design decisions
  2. ROS 2 Fluency: Design distributed robotics systems with ROS 2
  3. Simulation Mastery: Build high-fidelity simulations in Gazebo/Unity/Isaac
  4. Perception & Navigation: Implement SLAM, sensor fusion, and autonomous navigation
  5. AI Integration: Chain vision-language-action models into executable robot behaviors
  6. Sim-to-Real Transfer: Deploy simulated systems to physical robots safely

🚀 Hackathon Challenge​

This textbook is being developed for the AI/Spec-Driven Book Creation hackathon challenge:

"Write a book using Docusaurus and deploy it to GitHub Pages using Spec-Kit Plus and Claude Code."

Key Features:

  • ✅ Systematic spec-driven development workflow
  • ✅ All decisions documented in ADRs
  • ✅ Complete development history in PHRs
  • ✅ Quality validated against Constitution
  • ✅ Deployed automatically via GitHub Actions

→ Explore the Methodology

🤝 Contributing​

Want to add a chapter or improve existing content? Follow the same Spec-Driven workflow:

  1. Run /sp.specify "Your chapter idea"
  2. Review and approve the generated spec
  3. Run /sp.plan to create architectural design
  4. Run /sp.tasks to generate implementation tasks
  5. Run /sp.implement to execute the tasks
  6. All changes automatically tracked in PHRs and ADRs!

→ [Full contributing guide coming soon]

📖 How to Use This Textbook​

For Students​

  • Work through chapters sequentially (Part I → II → III)
  • Complete hands-on labs with provided code examples
  • Validate learning with quizzes and assessments
  • Build toward capstone: voice-commanded humanoid butler

For Instructors​

  • Use specs and plans as teaching guides
  • Reference ADRs to explain design decisions
  • Adapt content using same /sp.* workflow
  • Track student progress with assessment rubrics

For Researchers​

  • Review methodology pages for AI-augmented authoring insights
  • Examine PHRs for reproducibility and process transparency
  • Study ADRs for architectural decision patterns
  • Contribute improvements via spec-driven workflow

🔗 Quick Links​

Ready to begin? Start with Chapter 1: Introduction to Physical AI →

Or explore the Spec-Driven Workflow to understand how this book was built →

Built with ❤️ using Spec-Kit Plus, Claude Code, and Docusaurus