Skip to main content

Physical AI & Humanoid Robotics

Welcome to the AI-native textbook for the Physical AI & Humanoid Robotics course. This book blends theory, labs, and assessments with a built-in RAG chatbot, personalization, login via Better-Auth, and one-tap Roman Urdu translation.

What you will build​

  • A complete quarter-long pathway that bridges digital brains (LLMs, VLA) to physical bodies (ROS 2 humanoids, Gazebo/Isaac digital twins).
  • Capstone: a simulated humanoid that hears a voice command, plans with LLMs, navigates with Nav2/VSLAM, perceives with Isaac, and manipulates an object.
  • Cloud and on-prem lab patterns so students can work with either RTX rigs or cloud GPUs plus Jetson edge kits.

Interactive features baked into the book​

  • RAG Chatbot: Ask any question from the textbook or select text to ground the answer. Powered by OpenAI + Qdrant + Neon Postgres via the included FastAPI backend.
  • Login & Profile (Better-Auth): Sign up, capture hardware/software background, and sync learning state.
  • Personalize Chapter Content: A per-chapter button tailors explanations and lab hints to your profile.
  • Roman Urdu Toggle: Switch chapter text and chatbot responses to Roman Urdu instantly.

Course map (13 weeks)​

  1. Foundations of Physical AI and sensors 2-5. ROS 2: nodes, topics/services, launches, URDF for humanoids 6-7. Gazebo + Unity digital twins and sensor simulation 8-10. NVIDIA Isaac Sim/ROS for perception, RL, sim-to-real 11-12. Humanoid kinematics, bipedal balance, manipulation
  2. Vision-Language-Action: Whisper voice-to-action, GPT planning, multi-modal interaction

How to use this book​

  • Start at Plan for the build roadmap and architecture.
  • Follow Quickstart to run the FastAPI backend, Qdrant/Neon services, and the Docusaurus frontend locally.
  • Use the Chatbot widget (book sidebar) to stay grounded in the current page.
  • Press Personalize or Urdu at the top of any chapter once logged in.

Dive in, ship your capstone, and get ready to demo on Nov 30. 🚀