💻 computer science

Real-time Dynamics of Soft Manipulators with Cross-section Inflation: Application to the Octopus Muscular Hydrostat

This paper presents an extended Cosserat rod theory that incorporates in-plane inflation and viscoelastic effects to enable real-time, accurate dynamic modeling of soft manipulators, such as octopus arms, by adapting the Geometric Variable Strain approach for efficient numerical solutions.

Original authors: Yuchen Sun, Anup Teejo Mathew, Imran Afgan, Federico Renda, Cecilia Laschi

Published 2026-01-30

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Yuchen Sun, Anup Teejo Mathew, Imran Afgan, Federico Renda, Cecilia Laschi

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 are trying to build a robot arm that moves just like an octopus. Octopus arms are amazing: they can bend, twist, stretch, and even change how stiff they are, all without any hard bones or joints. They do this using "muscular hydrostats"—bundles of muscles that work together like a giant, flexible water balloon.

For a long time, scientists trying to model these soft robots used a mathematical tool called the Cosserat rod theory. Think of this classic theory like imagining the robot arm as a stack of rigid, flat coins glued together. You can bend the stack, twist it, or pull it apart, but the coins themselves never change shape. They stay perfectly flat and round.

The Problem:
Real octopus arms (and many soft robots) don't just bend; their cross-sections actually squish and expand. When an octopus wants to reach for food, it often squeezes its muscles sideways to make the arm longer and thinner, or squeezes it to make it stiffer. The old "stack of coins" model couldn't see this happening. It was like trying to describe a deflating balloon by only looking at a stack of rigid plates.

The Solution:
The authors of this paper created a new, upgraded version of the math. They added a special "inflation variable" (let's call it the squish-factor). Now, instead of just stacking rigid coins, their model treats the arm like a stack of rubber discs. These discs can stretch, twist, and—crucially—change their own size (inflate or deflate) as the arm moves.

Here is how they broke it down:

The "Squish" Factor: They added a new number to their equations that tracks how much the arm's cross-section is expanding or shrinking. This allows the model to capture the "active lateral traction" (the sideways squeezing) that real octopus muscles use.
The "Jelly" Physics: To make the math realistic, they combined two types of physics:
- Rubber-like behavior: How the material stretches and snaps back (hyperelasticity).
- Honey-like behavior: How the material resists moving quickly due to internal friction (viscosity).
- Analogy: Imagine pushing your hand through water (viscosity) while pulling a rubber band (elasticity). The model accounts for both.
Speeding it Up: Calculating the movement of a soft, squishy arm is usually very slow and heavy for computers. The authors used a clever trick called Geometric Variable Strain (GVS). Imagine instead of calculating every single atom in the arm, they described the arm's shape using a few smooth, wavy lines (like a musical note). This lets the computer solve the math in real-time, making it fast enough to control a robot while it's moving.

What They Tested:
They didn't just write the math; they tested it with three specific scenarios:

The Stiffness Test: They showed that by squeezing the arm sideways (using the new "squish" factor), the robot could resist being pushed down. The old model couldn't do this; it thought the arm would just collapse. The new model showed the arm getting rigid, just like a real octopus arm does when it needs to push against something.
The "Reaching" Test: They simulated an octopus arm reaching out. In nature, octopuses squeeze their muscles to make the arm longer and stiffer as it reaches. The new model successfully recreated this "stretching wave" and the changing thickness of the arm. The old model failed to show the arm getting longer; it just bent.
The "Fetching" Test: They simulated the arm grabbing something and pulling it back to the mouth. This involves twisting and bending so much that the tip might hit the body. The new model showed how the arm twists to avoid hitting itself, a movement that requires the cross-section to change shape.

The Bottom Line:
This paper gives us a better "instruction manual" for how soft, squishy robots move. By acknowledging that these robots can change their own thickness and stiffness (not just bend), the new math allows for much more realistic simulations. This helps engineers design better soft robots that can move with the same grace and adaptability as the octopus that inspired them.

Note: The paper focuses strictly on the mathematical modeling and simulation of these movements. It does not discuss clinical applications, medical uses, or future commercial products.

Technical Summary: Real-time Dynamics of Soft Manipulators with Cross-section Inflation

Problem Statement
Soft robotic arms, inspired by biological muscular hydrostats like the octopus, possess high dexterity and embodied intelligence, allowing them to perform complex tasks in unstructured environments. However, modeling their dynamics presents significant challenges. While classic Cosserat rod theory effectively models bending, twisting, shearing, and stretching of slender beams, it relies on the assumption of rigid cross-sections. This assumption fails to capture in-plane deformations, specifically cross-sectional inflation or contraction, which are critical for tasks involving active lateral traction (e.g., an octopus arm elongating via transversal muscle contraction). Existing reduced-order models often sacrifice accuracy for computational speed, while full 3D finite element methods (FEM) are computationally prohibitive for real-time control. Furthermore, previous extensions of Cosserat theory to include cross-sectional deformation have often been limited to static contexts or relied on simplified constitutive laws that deviate from physical reality.

Methodology
The authors propose an extended Cosserat rod theory that incorporates cross-sectional inflation as a dynamic variable. The methodology is structured as follows:

Kinematics and Strain Variables: The model introduces an additional strain variable, $\rho(s)$ , representing the cross-sectional inflation ratio of an axisymmetric geometry. This variable modifies the position vector of material points, allowing the cross-section to expand or contract homogeneously. The deformation gradient tensor is derived to include this inflation, leading to a modified Green-Lagrangian strain tensor.
Constitutive Law: To accurately describe the viscoelastic behavior of soft materials, the constitutive law integrates the Saint-Venant Kirchhoff (SVK) hyperelastic model with the Kelvin-Voigt viscous model. This combination accounts for both elastic energy storage and viscous dissipation. The authors derive explicit expressions for the strain energy density and dissipation rate, incorporating the new inflation variable and ensuring compatibility with incompressibility constraints (Poisson's ratio $\approx 0.5$ ).
Dynamics Formulation: The equations of motion are derived from the conservation of linear momentum, angular momentum, and material momentum. The material momentum balance yields a specific equation governing the evolution of the inflation ratio $\rho$ , coupled with the standard force and moment balance equations. External loads, including gravity, buoyancy, hydrodynamic drag, and added mass, are integrated into the model. Internal actuation loads are categorized into transversal (affecting cross-section size), longitudinal/oblique (affecting tension and bending), and point loads.
Numerical Solution (GVS): To enable real-time simulation, the authors adapt the Geometric Variable Strain (GVS) approach. The continuous strain fields (including the new inflation variable) are parameterized using polynomial basis functions (Legendre and Hermite polynomials) mapped to a finite set of generalized coordinates. This reduces the partial differential equations (PDEs) to a system of ordinary differential equations (ODEs), allowing for efficient recursive computation of kinematics, Jacobians, and dynamics.

Key Contributions

Extended Cosserat Theory: The derivation of a dynamic Cosserat rod theory that explicitly includes an in-plane inflation strain variable, addressing the geometric limitations of the classic theory regarding cross-sectional deformation.
Viscoelastic Constitutive Law: The development of a constitutive law combining SVK hyperelasticity and Kelvin-Voigt damping, providing a physically grounded description of soft material behavior under large deformations.
Reduced-Order Numerical Method: The adaptation of the GVS framework to the extended theory, enabling rapid and precise solutions for the dynamics of soft rods with variable cross-sections.
Comprehensive Load Modeling: A unified framework accounting for transversal, longitudinal, and oblique actuation loads, as well as complex environmental interactions (hydrodynamics).

Results and Applications
The paper validates the model through three simulation applications:

Axial Stiffness Tuning: The model demonstrates the ability to tune axial stiffness by modulating transversal actuation. Unlike the classic model, which cannot resist axial compression without buckling, the extended model shows that transversal contraction generates internal tension, allowing the manipulator to resist external compression forces and achieve rigid behavior. The simulated stiffness values align with analytical solutions.
Octopus Arm Reaching: The model simulates the reaching motion of an octopus arm, characterized by a wave of co-contraction of longitudinal and transversal muscles. The simulation reproduces the biological phenomenon of arm elongation driven by transversal muscle contraction. Comparisons with the classic Cosserat model and an extended model without transversal actuation show that the full model is necessary to capture the correct elongation and volume preservation (minimal volume change $\approx -2.99\%$ vs. significant loss in classic models).
Octopus Arm Fetching: The model simulates a fetching motion involving shortening and backward bending. It incorporates an oblique muscle to prevent self-collision during high-curvature bending. The simulation highlights the necessity of oblique actuation to avoid unrealistic self-intersection, a feature not captured by models lacking these specific actuation modes.

Significance and Claims
The paper claims that the proposed extended Cosserat rod model bridges the gap between the geometric accuracy required for complex biological motions and the computational efficiency needed for real-time control. By capturing in-plane inflation, the model offers a more realistic representation of muscular hydrostats, particularly regarding volume preservation and stiffness modulation. The authors assert that this framework enables the study of complex soft robotic motions, such as the reaching and fetching behaviors of octopuses, which were previously difficult to model accurately due to the limitations of rigid cross-section assumptions. The work provides a foundation for future research in the control and optimization of soft manipulators, leveraging the energy-based nature of the constitutive law for closed-loop control strategies. The authors acknowledge that the model remains a simplification, as it assumes axisymmetric cross-sections and does not fully enforce global volume invariance, suggesting future work could involve more general constitutive laws and strict volume constraints.

More like this