Skip to main content

Module 0: Foundations of Physical AI

Duration: Weeks 1 to 2 (2 weeks) Focus: Understanding embodied intelligence, physical AI principles, and the humanoid robotics landscape before diving into technical implementation

What You'll Build

By the end of this module, you will have:

  • Understanding of embodied intelligence and how it differs from digital AI
  • Knowledge of the humanoid robotics landscape and key players
  • Familiarity with sensor systems used in humanoid robots
  • Hands-on experience with basic robotics simulation using PyBullet
  • Foundation for understanding why ROS 2, Gazebo, and Isaac Sim matter

Module Project: Build a simple simulated robot in PyBullet that demonstrates basic physics understanding (gravity, collisions, sensor feedback).

Module Overview

Before we dive into ROS 2, Gazebo, and NVIDIA Isaac, it's crucial to understand why we're building humanoid robots and what makes Physical AI fundamentally different from the digital AI systems you may already know.

Physical AI represents the convergence of artificial intelligence with robotics—creating systems that don't just process information, but interact with the physical world. Unlike ChatGPT or image generators that exist purely in digital space, Physical AI systems must:

  • Understand physics: Gravity, friction, collisions, dynamics
  • Perceive the world: Process sensor data (cameras, LiDAR, IMUs)
  • Act in real-time: Make decisions and execute actions within milliseconds
  • Handle uncertainty: Deal with sensor noise, incomplete information, and unexpected events
  • Learn from interaction: Improve through trial and error in the real world

Why humanoids? Humanoid robots are designed to operate in human environments:

  • Navigate stairs, doors, and furniture designed for human bodies
  • Manipulate tools and objects designed for human hands
  • Collaborate with humans in shared workspaces
  • Learn from human demonstrations and interactions

This module sets the stage for everything that follows by establishing the conceptual foundation.

Learning Path

Chapter 0.1: Introduction to Embodied Intelligence & Physical AI

  • What is embodied intelligence?
  • Physical AI vs. digital AI: fundamental differences
  • Why robots need to understand physics
  • The role of simulation in Physical AI development

Chapter 0.2: Humanoid Robotics Landscape & Applications

  • History and evolution of humanoid robotics
  • Key players: Boston Dynamics, Tesla, Figure AI, Unitree
  • Current capabilities and limitations
  • Real-world applications and use cases
  • The path from simulation to deployment

Chapter 0.3: Sensor Systems for Humanoid Robots

  • Vision sensors: RGB cameras, depth cameras, stereo vision
  • LiDAR: 3D mapping and obstacle detection
  • IMUs: Balance, orientation, and motion tracking
  • Force/torque sensors: Tactile feedback and manipulation
  • Sensor fusion: Combining multiple modalities

Tools & Technologies

You will use:

  • PyBullet: Physics simulation engine - GitHub
  • Python 3.10+: Primary development language
  • NumPy: Numerical computing
  • Matplotlib: Visualization and plotting

Installation guides provided in Chapter 0.1.

Prerequisites

Required:

  • Programming: Basic Python knowledge (variables, functions, classes)
  • Mathematics: High school algebra and geometry
  • Physics: Basic understanding of forces, motion, gravity

No prior robotics experience needed—this module is designed as an entry point!

Week-by-Week Timeline

Week 1: Concepts & Landscape

  • Chapter 0.1: Introduction to Embodied Intelligence & Physical AI
  • Chapter 0.2: Humanoid Robotics Landscape & Applications

Week 2: Sensors & Hands-On

  • Chapter 0.3: Sensor Systems for Humanoid Robots
  • Module Project: PyBullet simulation project

Assessment

Weight: 10% of final grade - Module Project

Project: Basic Robot Simulation in PyBullet

Requirements:

  1. Functional:

    • Create a simple robot model (2 to 4 links)
    • Simulate gravity and collisions
    • Add at least one sensor (camera or IMU)
    • Demonstrate sensor feedback
  2. Technical:

    • Python script with PyBullet
    • Clear code comments
    • README with setup instructions
    • Screenshot or video of simulation running

Deliverables:

  • GitHub Repository:
    • /src: Python script(s)
    • /README.md: Setup and usage instructions
    • /screenshot.png or /demo_video.mp4: Simulation demonstration

Grading Rubric:

CriterionExcellent (90 to 100%)Good (75 to 89%)Needs Work (less than 75%)
FunctionalityAll features working, smooth simulationMost features workingMissing features or errors
Code QualityClean, well-commented, organizedReadable, some commentsHard to understand
DocumentationComplete setup guide, clear explanationsBasic instructionsMissing key info
UnderstandingDemonstrates grasp of physics conceptsShows basic understandingLimited understanding

Submission: Submit via course LMS by end of Week 2. Late penalty: -10% per day (max 2 days late).


Real-World Applications

What you'll understand after this module:

Industrial Humanoids:

  • Tesla Optimus: General-purpose humanoid for manufacturing and logistics
  • Figure AI: Humanoid robots for warehouse automation
  • Agility Robotics Digit: Last-mile delivery and logistics

Research Platforms:

  • Boston Dynamics Atlas: Advanced research in dynamic locomotion
  • Unitree G1: Open-source humanoid platform for research
  • NVIDIA Project GR00T: Foundation model for humanoid robots

Service Robots:

  • Pepper: Social interaction and customer service
  • NAO: Education and research platform
  • Sanbot: Healthcare and hospitality applications

Success Stories: What Students Built

Week 1 Milestone: Understanding of Physical AI principles and humanoid landscape

Week 2 Milestone: Working PyBullet simulation with sensors—ready for ROS 2 in Module 1!


Why This Module Matters

Without this foundation, you might:

  • Struggle to understand why ROS 2's architecture matters
  • Miss the importance of sensor fusion
  • Not appreciate the challenges of sim-to-real transfer
  • Lack context for why certain tools (Gazebo, Isaac Sim) exist

With this foundation, you'll:

  • Understand the "why" behind every technical decision
  • Appreciate the complexity of Physical AI systems
  • See how all modules connect to real-world deployment
  • Be motivated by the exciting applications ahead

Getting Help

Stuck on concepts?

Office Hours: See course schedule for TA support


Ready to Start?

This module provides the conceptual foundation for everything that follows. While it's lighter on code than later modules, the concepts here are crucial for understanding why we make certain technical choices and how Physical AI systems differ from digital AI.

Let's begin your journey into Physical AI.


Next: [Chapter 0.1: Introduction to Embodied Intelligence & Physical AI →](chapter-0 to 1.md)