Back to Portfolio
Screw Theory · Fundamentals

Introduction to Screw Theory

The elegant mathematics behind robot motion, explained with intuitive analogies

📖 20 min read 📐 Beginner friendly 🤖 Serial & Parallel Robots

Welcome! If you've ever wondered how roboticists describe robot motion mathematically, you're in the right place. Screw theory is one of the most powerful frameworks for understanding how robots move.

🎯 What You'll Learn

By the end of this tutorial, you'll understand what a "screw" is in robotics, how to describe any robot motion using twists, and why this framework is particularly powerful for analyzing parallel robots.

1. Why Screw Theory?

In robotics, we need to describe two fundamental things: how robots move (kinematics) and what forces they apply (statics). Traditional methods work, but screw theory offers something better.

💡 Think of it this way

Imagine describing how a door opens. You could track every point's coordinates—tedious! Or simply say: "it rotates around its hinges." Screw theory gives this elegant, geometric description for any rigid body motion.

Traditional Methods Screw Theory
Coordinate-dependent Coordinate-free (geometric)
Separate rotation & translation Unified description
Complex for parallel robots Natural for parallel robots
Singularities hard to visualize Geometric meaning

2. The Screw Analogy

The name comes from a beautiful observation: the most general motion of a rigid body is like a screw moving through wood—it rotates and translates simultaneously along the same axis.

θ (rotation) d (translation) Pitch h = d / θ h = 0 → Pure rotation h = ∞ → Pure translation 0 < h < ∞ → Helical

Figure 1: A screw motion combines rotation (θ) and translation (d) along the same axis. The pitch (h) determines the ratio.

🔩 Real-World Analogy

Think about screwing in a bolt: as you rotate it, it advances forward. The pitch determines how far it moves per rotation. A fine-thread screw has small pitch; a coarse-thread has larger pitch. We use this same concept to describe any rigid body motion!

Three Types of Motion

Pure Rotation pitch h = 0 Pure Translation pitch h = ∞ Helical Motion 0 < h < ∞

Figure 2: Every rigid body motion is either pure rotation, pure translation, or helical.

3. The Mathematics

Now let's put this intuition into mathematical form. Don't worry—we'll build it step by step!

📐 Definition: Screw

A screw is defined by:

We represent a screw as a 6-dimensional vector:

$$\mathcal{S} = \begin{pmatrix} \hat{s} \\ s_0 \end{pmatrix} = \begin{pmatrix} \hat{s} \\ q \times \hat{s} + h\hat{s} \end{pmatrix}$$

🎯 Why 6 Dimensions?

A rigid body in 3D has 6 degrees of freedom: it can translate in 3 directions (x, y, z) and rotate around 3 axes. So we need 6 numbers to fully describe its motion!

ŝ s₀ ← Direction (3D) ← Moment (3D) [ ] 𝒮 = ∈ ℝ⁶

Figure 3: A screw is a 6D vector with direction (top) and moment (bottom).

4. Twists: Robot Velocity

A twist is a special screw that describes velocity. It combines angular velocity and linear velocity into one elegant object.

📐 Definition: Twist

A twist $\xi$ describes instantaneous velocity:

$$\xi = \begin{pmatrix} \omega \\ v \end{pmatrix} \in \mathbb{R}^6$$

Where $\omega$ is angular velocity and $v$ is linear velocity.

O v (linear) ω (angular) TWIST ξ = [ω, v]ᵀ ω ∈ ℝ³ (angular velocity) v ∈ ℝ³ (linear velocity)

Figure 4: A twist combines angular velocity (ω) and linear velocity (v).

🚗 Car Analogy

Think of a car driving in a circle. At any instant, it has angular velocity (how fast it's turning) and linear velocity (how fast it's moving forward). The twist captures both in one object!

5. Wrenches: Forces

Just as twists describe velocity, wrenches describe forces and moments. They're the "dual" of twists.

📐 Definition: Wrench

A wrench $\mathcal{W}$ describes forces and moments:

$$\mathcal{W} = \begin{pmatrix} m \\ f \end{pmatrix} \in \mathbb{R}^6$$

Where $m$ is the moment (torque) and $f$ is the force.

The Power Equation

Twists and wrenches have a beautiful relationship. The power transmitted is their inner product:

$$P = \xi^T \mathcal{W} = \omega \cdot m + v \cdot f$$

⚡ Key Insight

Power = rotational power ($\omega \cdot m$) + translational power ($v \cdot f$). This elegant formula is fundamental to robot statics and dynamics!

6. Serial Robots

For a serial robot (like a typical robot arm), each joint contributes a twist to the end-effector motion. The Jacobian matrix collects these twists.

1 2 3 EE ξ₁ ξ₂ ξ₃ Jacobian Matrix J = [ξ₁ | ξ₂ | ξ₃] ξₑₑ = J · q̇ 6×3 matrix

Figure 5: A 3R serial robot. Joint twists form the Jacobian columns.

🔑 Key Formula

The Jacobian is formed by stacking joint twists as columns:

$$J = \begin{bmatrix} \xi_1 & \xi_2 & \cdots & \xi_n \end{bmatrix}$$

End-effector velocity: $\xi_{ee} = J \dot{q}$

Joint Twist Formulas

Joint Type Motion Unit Twist
Revolute Rotation about axis $\hat{s}$ through $q$ $\xi = [\hat{s},\; q \times \hat{s}]^T$
Prismatic Translation along $\hat{s}$ $\xi = [0,\; \hat{s}]^T$

7. Parallel Robots

Screw theory truly shines with parallel robots! Multiple chains connect the base to the platform, creating constraints.

Base (Fixed) Platform (Moving) Chain 1 Chain 2 Chain 3 ξₚ KEY CONSTRAINT All chains → same platform twist

Figure 6: A parallel robot. Each chain constrains platform motion.

🔑 Constraint Equation

For parallel robots:

$$A \dot{x} = B \dot{q}$$

$A$ captures platform constraints, $B$ captures actuator effects. The Jacobian is $J = A^{-1}B$.

⚠️ Singularities

A singularity occurs when $\det(A) = 0$ or $\det(B) = 0$. The robot loses control or mobility. Screw theory reveals their geometric meaning!

8. Example: 2R Planar Robot

Let's work through a complete example with a 2R planar robot.

y x Origin θ₁ L₁ θ₂ A L₂ P (EE) z (out) Given: L₁ = 1.0 m L₂ = 0.8 m θ₁ = 45° θ₂ = 30°

Figure 7: 2R planar robot with links L₁, L₂ and angles θ₁, θ₂.

Step-by-Step Solution

Identify Joint Axes

Both joints are revolute with axes perpendicular to the plane (pointing in $+z$). So: $\hat{s} = [0, 0, 1]^T$ for both joints.

Find Joint Positions

Joint 1 at origin: $q_1 = [0, 0, 0]^T$
Joint 2 at point A: $q_2 = [L_1 \cos\theta_1,\; L_1 \sin\theta_1,\; 0]^T$

Compute Unit Twists

Using $\xi = [\hat{s},\; q \times \hat{s}]^T$ for revolute joints:

$$\xi_1 = \begin{pmatrix} 0 \\ 0 \\ 1 \\ 0 \\ 0 \\ 0 \end{pmatrix}, \quad \xi_2 = \begin{pmatrix} 0 \\ 0 \\ 1 \\ -L_1 \sin\theta_1 \\ L_1 \cos\theta_1 \\ 0 \end{pmatrix}$$

Build the Jacobian

Stack twists as columns:

$$J = \begin{pmatrix} 0 & 0 \\ 0 & 0 \\ 1 & 1 \\ 0 & -L_1 s_1 \\ 0 & L_1 c_1 \\ 0 & 0 \end{pmatrix}$$

Where $s_1 = \sin\theta_1$, $c_1 = \cos\theta_1$.

Compute End-Effector Velocity

Given joint velocities $\dot{\theta}_1$ and $\dot{\theta}_2$:

$$\xi_{EE} = J \begin{pmatrix} \dot{\theta}_1 \\ \dot{\theta}_2 \end{pmatrix}$$

✅ Numerical Result

With $\dot{\theta}_1 = 1$ rad/s, $\dot{\theta}_2 = 0.5$ rad/s:
Angular velocity: $\omega_z = 1 + 0.5 = 1.5$ rad/s
The Jacobian maps joint velocities to end-effector velocity!

9. Summary

📝 Key Takeaways

Screw theory provides a coordinate-free, geometric way to understand robot motion. Once you internalize these concepts, analyzing complex mechanisms becomes intuitive!

🚀 Next Steps