⚡ electrical engineering

A Geometric Task-Space Port-Hamiltonian Formulation for Redundant Manipulators

This paper introduces a novel geometric port-Hamiltonian formulation for redundant manipulators that decomposes momentum into task-space and null-space variables, demonstrating its effectiveness through an IDA-PBC control design that stabilizes and shapes the impedance of a 7-DOF Emika Panda robot in simulation.

Original authors: Federico Califano, Camilla Rota, Riccardo Zanella, Antonio Franchi

Published 2026-06-23

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Federico Califano, Camilla Rota, Riccardo Zanella, Antonio Franchi

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you have a robotic arm with seven joints (like a human arm with an extra degree of freedom). This is called a "redundant" manipulator because it has more joints than strictly necessary to reach a specific point in space. If you want the robot's hand to move in a straight line, there are many different ways the individual joints can bend and twist to achieve that same hand movement.

This paper presents a new mathematical "language" (a geometric Port-Hamiltonian formulation) to describe how these redundant robots move and interact with the world. Here is the breakdown using simple analogies:

1. The Core Idea: Splitting the Robot's "Brain"

Usually, when engineers model a robot, they look at all the joints together as one big, messy knot. This new approach suggests we should split the robot's movement into two distinct "channels":

The Task Channel (The "What"): This is the part of the movement that actually moves the robot's hand to where you want it. Think of this as the "mission."
The Null Space Channel (The "How"): This is the extra movement the robot can do without moving its hand at all. Imagine a snake sliding its body while its head stays perfectly still; that sliding is the "null space."

The authors created a mathematical map that separates these two channels cleanly. They show that the robot's energy (kinetic energy) and the forces it uses can be split into a "Task Part" and a "Null Space Part" that don't interfere with each other.

2. The "Energy Bank" Analogy

In physics, robots store energy like a bank account.

Old Way: The bank account was a single lump sum. It was hard to tell how much energy was being used to move the hand versus how much was being wasted (or used) by the extra joints wiggling around.
New Way: The authors opened two separate bank accounts.
1. Task Account: Tracks energy used to move the hand.
2. Null Space Account: Tracks energy used by the extra joints.

This separation is crucial because it allows engineers to control the hand's movement (the Task) without accidentally messing up the internal posture of the robot, or vice versa.

3. The "Power Ports" (Doors for Energy)

The paper introduces the idea of "ports." Imagine the robot has two different doors where energy can flow in or out:

The Task Door: This is where the robot interacts with the outside world (e.g., pushing a box, holding a tool).
The Null Space Door: This is where the robot manages its own internal posture.

The authors' model shows that energy flowing through the Task Door is completely independent of energy flowing through the Null Space Door. This makes it much easier to design controllers (the robot's "brain") that can handle both tasks simultaneously without confusion.

4. The Control Strategy: "Shaping the Spring"

The paper doesn't just describe the robot; it uses this new model to build a controller called IDA-PBC. Think of this as a way to "program the robot's personality."

Stabilization: The controller ensures the robot stays steady at a specific spot.
Impedance Shaping: This is the cool part. The controller can make the robot's hand feel "soft" (like a marshmallow) or "stiff" (like a rock) when it touches something.
- Analogy: Imagine the robot's hand is a spring. The authors show how to mathematically tune that spring so that if you push the robot, it pushes back with exactly the right amount of force, whether you want it to be gentle or firm.

5. The Simulation Test

To prove this works, the authors simulated a 7-jointed robot (a Franka Emika Panda). They made the robot reach a specific point and then gave it a "push" (an external force).

They tested two different internal postures (different ways the joints could be arranged to reach the same point).
They tested two different "stiffness" settings (soft vs. stiff).
Result: In all cases, the robot stayed stable, returned to its target, and behaved exactly like the "spring" they programmed it to be. The energy in the system behaved predictably, proving their new mathematical model works.

Summary

In short, this paper gives engineers a better blueprint for building and controlling redundant robots. Instead of treating the robot as a confusing jumble of joints, it provides a clear way to separate the "goal" (moving the hand) from the "freedom" (how the joints wiggle). This allows for smarter, safer, and more adaptable robots that can be programmed to feel soft or stiff exactly how a human operator needs them to.

Technical Summary: A Geometric Task-Space Port-Hamiltonian Formulation for Redundant Manipulators

Problem Statement
Redundant manipulators, which possess more degrees of freedom (DOF) than required for a specific differential kinematic task, present a challenge in modeling and control. While Lagrangian formulations for redundancy are well-established in literature (e.g., Dietrich et al., 2015; Ott et al., 2015), there is a need for a rigorous geometric framework within the Port-Hamiltonian (pH) formalism that explicitly handles the decomposition of energy and power between the task space and the null space. Existing pH approaches for task-space impedance shaping (e.g., Scherpen et al., 2025) often address non-redundant systems or do not fully exploit the structural properties of redundancy. The core problem addressed is the lack of a pH model that naturally decomposes the system's momentum and power ports into task-space and null-space components, facilitating energy-based analysis and control design for redundant systems.

Methodology
The authors propose a novel geometric pH formulation derived from a coordinate transformation of the standard Hamiltonian dynamics. The methodology proceeds through the following steps:

Geometric Decomposition:
- Velocity Space: Using the inertia metric $M(q)$ , the tangent space $T_qQ$ is decomposed into the kernel of the task Jacobian $J(q)$ (null space) and its $M$ -orthogonal complement. This allows the joint velocity $\dot{q}$ to be uniquely split into a task-space component $v$ and a null-space component $\nu$ .
- Force Space: A dual decomposition is applied to the generalized force space $T^*_qQ$ . The torque $\tau$ is decomposed into a component $\tau_F$ in the image of $J^\top$ (task space forces) and a component $\tau_0$ in the annihilator of the task-space velocity subspace (null space forces).
- Power Balance: This decomposition reveals that the mechanical power $\tau^\top \dot{q}$ splits into two independent terms: $\tau_F^\top v$ (task power) and $\tau_0^\top \nu$ (null-space power).
Coordinate Transformation:
- To embed this decomposition into the pH structure, the authors define an extended task velocity $\eta_e = \bar{J}(q)\dot{q}$ , where $\bar{J}$ combines the task Jacobian $J$ and a null-space basis matrix $N$ .
- A change of coordinates $\Phi$ is applied to the standard Hamiltonian state $(q, p)$ , transforming the momentum $p$ into an extended momentum $\pi_e = \bar{J}^{-\top}(q)p$ .
- This transformation yields a new state vector $z = (q, \pi, \pi_\nu)$ , where $\pi$ represents the task-space momentum and $\pi_\nu$ the null-space momentum. The kinetic energy is shown to decompose into a sum of task-space and null-space quadratic forms involving the inverse of the extended mobility tensor $\bar{\Lambda}$ .
System Formulation:
- The resulting system is a port-Hamiltonian system (Equation 27) with a skew-symmetric interconnection matrix $J_z$ . The system features two distinct power ports: a task port $(\sigma, \eta)$ and a null-space port $(\tau_0, \nu)$ .
- The interconnection matrix includes off-diagonal terms ( $G_{12}$ ) that explicitly model the internal energy coupling between task and null-space dynamics.
Control Design (IDA-PBC):
- The authors apply the Interconnection and Damping Assignment Passivity-Based Control (IDA-PBC) framework to the new model.
- The control objective is to stabilize a desired task-space equilibrium $x^*$ and shape the task-space impedance.
- The matching conditions are solved by preserving the kinetic energy structure and introducing a shaped potential energy $V_c(q)$ .
- The resulting control law consists of three components: Energy Shaping (to define the equilibrium), Energy Routing (to modify interconnections without dissipating energy), and Damping Injection (to ensure asymptotic stability).

Key Contributions

Novel Geometric pH Model: The paper presents the first pH formulation for redundant manipulators that rigorously decomposes momentum and power into task-space and null-space components using a coordinate transformation.
Dual Decomposition: It introduces a dual decomposition of the generalized force space that is physically meaningful and mathematically consistent with the velocity decomposition, utilizing dynamically consistent projectors.
Explicit Power Ports: The model explicitly identifies two power-conjugated port pairs: one for the task space (modeling external interactions) and one for the null space (modeling internal redundancy dynamics).
Impedance Shaping via IDA-PBC: The authors demonstrate the utility of the model by designing an IDA-PBC controller that stabilizes a task-space configuration and actively shapes the impedance (stiffness and damping) at the task port.

Results
The proposed methodology was validated through simulations on a 7-DOF Franka Emika Panda robot performing a 3-DOF Cartesian position task.

Stability: The closed-loop system was shown to be stable, with the total Hamiltonian $H_{cl}$ remaining non-increasing in the absence of external perturbations.
Impedance Behavior: The simulations demonstrated the ability to shape the task-space impedance. By varying the potential energy term $V_c(q)$ , the authors achieved different stiffness profiles (soft vs. stiff) at the equilibrium.
Redundancy Handling: The results confirmed that the controller could stabilize the task-space position $x^*$ while managing the null-space dynamics. Different joint configurations $q^*$ leading to the same $x^*$ were successfully stabilized with distinct null-space energy behaviors, illustrating the controller's ability to handle the redundancy.
Energy Trajectories: The simulation plots showed the evolution of task-space energy ( $\bar{K}_t$ ), null-space energy ( $\bar{K}_\nu$ ), and total energy ( $H_{cl}$ ), confirming the expected energy exchange and dissipation characteristics.

Significance and Claims
The authors claim that this work provides a rigorous theoretical foundation for analyzing and controlling redundant manipulators within the port-Hamiltonian framework. By decomposing the system into task and null-space ports, the model facilitates:

Energy-Based Analysis: It simplifies the analysis of energy flows and passivity properties compared to traditional Lagrangian descriptions.
Modular Control: It opens new possibilities for task-space control strategies that explicitly account for redundancy, allowing for the independent shaping of task impedance and null-space behavior.
Structural Insight: The formulation reveals the geometric nature of the coupling between task and null-space dynamics, which is often obscured in other formulations.

The paper concludes that this formulation is a significant step toward developing advanced, energy-aware control strategies for redundant robotic systems, particularly for applications requiring precise impedance control and safe physical interaction.