Skip to main content

Chapter 2: Humanoid Robotics Landscape

Overview​

What You'll Learn​

  • Identify major humanoid robot platforms and their capabilities
  • Explain embodiment principles and morphological considerations
  • Analyze kinematic structures and degrees of freedom
  • Compare design philosophies across different platforms

Prerequisites​

  • ✅ Completed Chapter 1
  • ✅ Understanding of Physical AI concepts

Chapter Roadmap​

Covers modern humanoid platforms (Tesla, Unitree, Boston Dynamics), embodiment theory, kinematics, and hardware-software co-design.

Estimated Time: 4-5 hours

Background​

The humanoid form factor has evolved from research curiosity to commercial viability. Modern platforms leverage AI, improved actuators, and cost reduction to create capable general-purpose robots.

Evolution of Humanoid Robotics​

1970s-1990s: WABOT-1 (1973) first full-scale humanoid, Honda P-series research
2000-2010: ASIMO, HRP series, research focus
2010-2020: Atlas, HUBO, Pepper - demonstrations but limited deployment
2020+: Optimus, G1, Figure 01 - commercial deployment focus

Core Concepts​

Concept 1: Modern Humanoid Platforms​

Tesla Optimus (Tesla Bot)​

  • Purpose: General-purpose assistant, manufacturing automation
  • Height: ~5'8" (173 cm), Weight: ~125 lbs (57 kg)
  • Actuators: 40+ DOF, custom linear actuators
  • Sensors: 8 cameras (FSD computer), IMU
  • AI: End-to-end neural networks, Vision-Language-Action
  • Status: Prototypes in Tesla factories (2024)

Unitree G1/H1​

  • Purpose: Research platform, affordable humanoid
  • Cost: G1 ~$16,000 (breakthrough pricing)
  • DOF: 23-27 degrees of freedom
  • Sensors: 3D LiDAR, depth cameras, joint encoders
  • Strengths: Open platform, good value
  • Applications: Research, education, light industrial

Boston Dynamics Atlas​

  • Purpose: Research platform, dynamic locomotion
  • Capabilities: Parkour, backflips, complex manipulation
  • Hydraulic: Earlier versions, newer models electric
  • Control: Model-based control + RL
  • Status: Not commercialized, research demonstrations

Agility Robotics Digit​

  • Purpose: Logistics, warehouse automation
  • Design: Torso-legs only (no arms initially, arms added later)
  • Deployments: Amazon warehouses, Ford plants
  • Focus: Reliability over versatility

Concept 2: Embodiment and Morphology​

Morphology = physical form and structure

Why Humanoid?​

  • Human environments designed for human dimensions
  • Stairs, doors, handles assume human morphology
  • Can use human tools without redesign
  • Leverage human demonstration data

Trade-offs:​

Advantages:

  • General purpose (one platform, many tasks)
  • Intuitive for humans to work alongside
  • Large training dataset (human videos)

Disadvantages:

  • Unstable (bipedal balancing complex)
  • Expensive (many actuators)
  • Not optimal for any single task

Concept 3: Kinematics and Degrees of Freedom​

Degrees of Freedom (DOF): Number of independent ways robot can move

Typical Humanoid DOF Budget:​

  • Legs: 6 DOF per leg (hip: 3, knee: 1, ankle: 2) = 12 total
  • Arms: 7 DOF per arm (shoulder: 3, elbow: 1, wrist: 3) = 14 total
  • Hands: 12-20 DOF per hand = 24-40 total
  • Torso/Neck: 3-6 DOF
  • Total: 50-70 DOF for full humanoid

More DOF = more versatility but harder to control.

Figure 2.1: Typical humanoid DOF distribution

Implementation​

Tutorial: Analyzing Platform Specifications​

Compare three platforms across key dimensions:

# Humanoid comparison framework
platforms = {
    "Tesla Optimus": {
        "dof": 40,
        "weight_kg": 57,
        "height_m": 1.73,
        "cost_est": "Unknown",
        "status": "Development",
        "strength": "AI integration"
    },
    "Unitree G1": {
        "dof": 23,
        "weight_kg": 35,
        "height_m": 1.27,
        "cost_est": "$16,000",
        "status": "Available",
        "strength": "Affordability"
    },
    "Boston Dynamics Atlas": {
        "dof": 28,
        "weight_kg": 89,
        "height_m": 1.5,
        "cost_est": "$150,000+",
        "status": "Research",
        "strength": "Agility"
    }
}

Lab Exercises​

Lab 1: Platform Comparison Analysis​

Objective: Create detailed comparison of humanoid platforms

Deliverables:

  • Comparison table (DOF, sensors, capabilities)
  • Analysis of design philosophies
  • Recommendation for specific use cases

Summary​

Key Takeaways​

  1. Modern humanoids span from $16K (Unitree) to $150K+ (Atlas) with varying capabilities
  2. Humanoid form chosen for versatility in human environments, not task optimization
  3. DOF budget determines versatility vs. control complexity
  4. Design philosophy varies: AI-first (Tesla) vs. dynamics-first (Boston Dynamics)

Next Steps​

➡️ Chapter 3: Sensor Foundations