Skip to main content

Module 08: Locomotion

Introduction​

Bipedal locomotion is perhaps the most challenging aspect of humanoid robotics. This module covers the dynamics, stability analysis, and control strategies for walking robots.

Section 1: Stability Criteria​

1.1 Zero Moment Point​

Zero Moment Point (ZMP): The point on the ground where the sum of moments due to gravity and inertial forces equals zero. When ZMP is within the support polygon, the robot is dynamically stable.

pZMP=βˆ‘imi(zΒ¨i+g)piβˆ’βˆ‘imipΒ¨iziβˆ‘imi(zΒ¨i+g)\mathbf{p}_{ZMP} = \frac{\sum_i m_i(\ddot{z}_i + g)p_i - \sum_i m_i\ddot{p}_i z_i}{\sum_i m_i(\ddot{z}_i + g)}

1.2 Support Polygon​

The convex hull of contact points defines stable ZMP locations.

Section 2: Gait Generation​

2.1 Central Pattern Generators​

Neural-inspired oscillators for rhythmic movement:

class CPGOscillator:
def __init__(self, frequency, amplitude):
self.freq = frequency
self.amp = amplitude
self.phase = 0

def step(self, dt):
self.phase += 2 * np.pi * self.freq * dt
return self.amp * np.sin(self.phase)

2.2 Preview Control​

Using ZMP trajectory preview for stable walking.

ZMP-based control assumes flat ground. On uneven terrain, additional measures (foot placement adaptation, compliant control) are needed.

Section 3: Whole-Body Control​

3.1 Task-Space Inverse Dynamics​

Hierarchical control framework:

Ο„=MqΒ¨d+hβˆ’JcTfc\boldsymbol{\tau} = \mathbf{M}\ddot{\mathbf{q}}_d + \mathbf{h} - \mathbf{J}_c^T\mathbf{f}_c

3.2 Operational Space Control​

Decoupling tasks in Cartesian space.

Section 4: Capture Point Dynamics​

4.1 Linear Inverted Pendulum Model​

Simplified dynamics for walking:

Capture Point: The location where the robot must step to come to rest without falling. Enables push recovery and dynamic balance.

4.2 Step Planning​

Real-time foot placement based on capture point.

Summary​

Key takeaways:

  1. ZMP within support polygon ensures stability
  2. Gait generation can use CPGs or optimization
  3. Whole-body control coordinates all DOFs
  4. Capture point enables reactive balance control

Key Concepts​

  • ZMP: Zero Moment Point stability criterion
  • Support Polygon: Convex hull of contact points
  • Whole-Body Control: Coordinating all robot DOFs
  • Capture Point: Reactive balance metric

Further Reading​

  1. Vukobratovic, M. & Borovac, B. (2004). "Zero-Moment Pointβ€”Thirty Five Years of Its Life"
  2. Kajita, S. et al. (2014). "Introduction to Humanoid Robotics"