The Jacobian matrix is the fundamental tool for analyzing robot velocity, force transmission, and singularities. In parallel robots, understanding the Jacobian is essential because it reveals where the robot loses control or stiffness.
💡 WHAT IS THE JACOBIAN?
The Jacobian is a matrix that maps joint velocities (q̇) to end-effector velocities (ẋ). It tells us how fast the end-effector moves for a given joint motion.
1. The Velocity Equation
For any robot, the relationship between joint velocities and end-effector velocity is given by:
ẋ = J · q̇
End-effector velocity = Jacobian × Joint velocities
Where:
- ẋ ∈ ℝm is the end-effector velocity (twist)
- J ∈ ℝm×n is the Jacobian matrix
- q̇ ∈ ℝn is the joint velocity vector
2. Serial vs Parallel Robot Jacobians
The key difference between serial and parallel robots lies in how we formulate the Jacobian:
Serial robots have one Jacobian; parallel robots typically require two
Serial Robot Jacobian
ẋ = J · q̇
Single Jacobian J relates joint velocities to end-effector
Parallel Robot Jacobian
Jx · ẋ = Jq · q̇
Two Jacobians: Jx (end-effector side) and Jq (actuator side)
Rearranging gives us the overall Jacobian:
ẋ = Jx-1 · Jq · q̇ = J · q̇
Where J = Jx-1 · Jq
3. Example: 3-RRR Planar Parallel Robot
Let's analyze one of the most common parallel robots: the 3-RRR planar parallel manipulator. It has 3 degrees of freedom (x, y, θ) and consists of three identical RRR legs.
3-RRR planar parallel robot: 3 actuated joints (θ₁, θ₂, θ₃) control end-effector position (x, y) and orientation (θ)
Deriving the Jacobian
Write the loop closure equations
For each leg, the vector from base to platform must close:
Ai + L₁·û1i + L₂·û2i = P + R(θ)·Ci
For i = 1, 2, 3 (each leg)
Differentiate with respect to time
Taking the time derivative gives us velocity relationships:
L₁·θ̇1i·n̂1i + L₂·θ̇2i·n̂2i = ẋ
Where n̂ is the unit vector perpendicular to the link
Eliminate passive joint velocities
Dot both sides with û2i to eliminate θ̇2i:
L₁·sin(θ2i - θ1i)·θ̇1i = û2i · ẋ
Assemble the Jacobian matrices
Writing in matrix form for all three legs:
Jq · q̇ = Jx · ẋ
3×3 diagonal actuator Jacobian and 3×3 end-effector Jacobian
4. Singularities: When Things Go Wrong
Singularities occur when the Jacobian loses rank — meaning the robot loses the ability to move in certain directions or resist forces.
⚠️ WARNING: SINGULARITIES ARE DANGEROUS
At a singularity, the robot may:
- Lose control of certain motions
- Experience infinite forces in actuators
- Gain unwanted degrees of freedom
- Become unstable or unpredictable
Types of Singularities in Parallel Robots
| Type | Condition | Effect |
|---|---|---|
| Type I (Serial) | det(Jq) = 0 | Robot at workspace boundary; can't move in some directions |
| Type II (Parallel) | det(Jx) = 0 | Gains DOF; platform can move with locked actuators |
| Type III (Combined) | Both = 0 | Most dangerous; complex behavior |
Singularities occur when legs align or fully extend — avoid these configurations!
5. Practical Applications
Velocity Analysis
Given desired end-effector velocity ẋ, find required joint velocities:
q̇ = J-1 · ẋ
Inverse velocity kinematics
Force Analysis (Statics)
The Jacobian transpose relates end-effector forces to joint torques:
τ = JT · F
Where τ = joint torques, F = end-effector wrench
Manipulability
The manipulability index measures how well the robot can move:
w = √det(J · JT)
w = 0 at singularities; maximize w for best performance
🎯 DESIGN TIP
When designing a parallel robot, always analyze the Jacobian throughout the workspace. Ensure singularities are outside the desired working area, and maximize manipulability in the regions where the robot will operate most.
Summary
- The Jacobian maps joint velocities to end-effector velocities
- Parallel robots have two Jacobians: Jx and Jq
- Type I singularities: workspace boundary (det(Jq) = 0)
- Type II singularities: uncontrolled motion (det(Jx) = 0)
- The Jacobian transpose relates forces: τ = JT · F
- Manipulability measures motion capability; zero at singularities