Linear viscoelastic rheological FrBD models

This paper introduces two novel rate-and-state-dependent friction models within the Friction with Bristle Dynamics (FrBD) framework based on Generalized Maxwell and Generalized Kelvin-Voigt viscoelastic elements, demonstrating that they satisfy boundedness and passivity for any physically meaningful parameters and illustrating their utility for control design in robotics.

The Gist

Imagine you are trying to push a heavy box across a carpet. You don't just push, and it doesn't just slide. There's a moment where you push hard, but the box doesn't move yet (it's "stuck"). Then, suddenly, it jerks forward. Once it's moving, it feels easier to keep it going, but if you stop and start again, it's hard to get it moving.

This complex "stick-slip" behavior is friction. Engineers have been trying to write a perfect mathematical recipe for friction for decades because it's crucial for robots, car brakes, and even your phone's touchscreen.

This paper introduces a new, smarter way to model that friction. Here is the breakdown using simple analogies.

1. The Old Way: The "Stiff Spring" Problem

For a long time, the most popular model (called LuGre) imagined friction as a single, stiff spring covered in tiny bristles (like a toothbrush). When you push, the bristles bend.

• The Problem: This model was great at predicting movement, but it had a fatal flaw for safety-critical systems like robots: it wasn't "passive."
• The Analogy: Imagine a robot arm holding a glass of water. If the friction model is "non-passive," it's like the robot arm suddenly deciding to add energy to the system out of nowhere. The arm might start shaking violently, spilling the water, because the math told it to push harder than physics allowed. It's like a car that accelerates on its own when you hit the brakes.

2. The New Idea: "Friction with Bristle Dynamics" (FrBD)

The authors propose a new framework called FrBD. Instead of just one stiff spring, they imagine the surface is made of viscoelastic materials (like silly putty or a memory foam mattress). These materials have two superpowers:

1. Elasticity: They bounce back (like a spring).
2. Viscosity: They flow slowly and resist motion (like honey).

The paper introduces two specific "recipes" for these materials, based on classic physics models:

• The Generalized Maxwell (GM) Model: Think of this as a bunch of springs and shock absorbers connected in a row. If you pull one end, the whole chain stretches and slowly settles.
• The Generalized Kelvin-Voigt (GKV) Model: Think of this as springs and shock absorbers connected side-by-side. If you push, they all resist together, but some parts move faster than others.

3. Why This is a Big Deal

The authors proved two massive things about these new models:

A. They are "Safe" (Passive)
In the world of control theory, "passive" means the system never creates energy out of thin air; it only absorbs or stores it.

• The Analogy: Think of a passive friction model as a sponge. If you squeeze it, it absorbs your energy. It never suddenly shoots water back at you with more force than you gave it.
• The Result: Because these new models are guaranteed to be passive, engineers can use them to design robot controllers that are mathematically guaranteed not to go crazy or become unstable.

B. They are "Realistic" (Bounded)
The models proved that no matter how you push or pull, the forces and internal movements will never go to infinity.

• The Analogy: It's like a car suspension. No matter how big a bump you hit, the car doesn't fly into space; the suspension absorbs the shock and stays within a safe range.

4. What Does It Actually Do?

The paper shows that these models can reproduce two tricky behaviors that older models struggled with:

• Frictional Lag (The "Hysteresis" Loop):

  • Scenario: You speed up a car, then slow down. The friction isn't the same at the same speed in both directions.
  • The Model: The new models act like a rubber band. If you stretch it fast, it feels different than if you stretch it slowly. The model captures this "memory" of how fast you are moving, creating a realistic loop in the data.

• Relaxation (The "Settling" Effect):

  • Scenario: You push a heavy object, and it starts moving. The friction force doesn't instantly settle to a steady number; it takes a moment to "relax" into place.
  • The Model: Because the model uses multiple "shock absorbers" (the Generalized parts), it can mimic materials that have different relaxation speeds, just like how a thick gel takes longer to settle than a thin liquid.

5. The Real-World Test: The Robot Arm

To prove it works, the authors simulated a robotic arm.

• They used their new "safe" friction model to build a controller.
• Because the model was passive, the controller could easily "talk" to the robot without fighting it.
• The Outcome: The robot arm tracked its target perfectly and didn't shake or overshoot, even with complex friction.

Summary

This paper is like upgrading the operating system for robot friction.

• Old System: A simple, sometimes glitchy spring that could make robots unstable.
• New System: A sophisticated, multi-layered "silly putty" model that is mathematically proven to be safe, stable, and incredibly realistic.

It allows engineers to build robots that move more smoothly, handle delicate objects better, and are safer to work around, all by treating friction not as a simple brake, but as a complex, energy-absorbing material.

Technical Summary

Here is a detailed technical summary of the paper "Linear viscoelastic rheological FrBD models" by Romano et al.

1. Problem Statement

Friction is a critical factor in the dynamics and control of mechanical systems, particularly in high-precision robotics and mechatronics. Existing dynamic friction models, such as the widely used LuGre model, suffer from significant limitations:

• Lack of Passivity: The LuGre model is not inherently passive for general parameter choices, which complicates stability analysis and robust control design.
• Physical Interpretability: Many parameters in existing models lack clear physical meaning, hindering their application in multibody simulations.
• Limited Rheological Scope: While rate-and-state models capture complex phenomena (hysteresis, memory, pre-sliding), they often fail to systematically integrate with classical linear viscoelasticity theories (e.g., Generalized Maxwell or Kelvin-Voigt models) used to describe solid materials like polymers and rubbers.

The paper aims to bridge this gap by developing a new class of friction models that are rate-and-state-dependent, inherently passive, and grounded in linear viscoelastic rheology.

2. Methodology

The authors build upon the Friction with Bristle Dynamics (FrBD) framework, a paradigm that combines nonlinear friction laws with constitutive equations for internal "bristle" elements. The methodology involves three main steps:

1. Rheological Formulation:
   The authors introduce two novel formulations based on the two most general linear viscoelastic models for solids:

   • Generalized Maxwell (GM): Represents the friction force as a sum of a spring and $n$ parallel branches, each containing a spring and a dashpot in series.
   • Generalized Kelvin-Voigt (GKV): Represents the friction force as a spring in parallel with $n$ branches, where each branch contains a spring and a dashpot in series.

2. Model Derivation:

   • The bristle deflection $z$ generates a force $f$ governed by the chosen constitutive equation (GM or GKV).
   • The relative sliding velocity at the bristle tip is defined as $v_s = v - \dot{z}$, where $v$ is the relative velocity of the bodies.
   • By equating the constitutive force to the friction law $f_r(v_s) = \text{sgn}(v_s)\mu(v_s)|v_s|_\epsilon$, the authors derive a system of $n+1$ nonlinear Ordinary Differential Equations (ODEs) for both GM and GKV variants.
   • The derivation utilizes the Implicit Function Theorem (Theorem 1.1) to ensure the existence of a unique solution for the bristle velocity $\dot{z}$.

3. Theoretical Analysis:
   The resulting models are rigorously analyzed for:

   • Well-posedness: Existence and uniqueness of solutions.
   • Boundedness: Ensuring states and forces remain finite.
   • Passivity: Proving that the energy dissipated by friction is non-negative, a crucial property for control stability.

3. Key Contributions

• Novel FrBD Formulations: The paper introduces the FrBD$_{n+1}$-GM and FrBD$_{n+1}$-GKV models. These generalize previous models (like LuGre and FrBD$_1$-KV) by incorporating multiple relaxation times, allowing for richer descriptions of viscoelastic materials.
• Guaranteed Passivity: Unlike the LuGre model, which requires specific damping terms to be passive, the proposed FrBD models are inherently passive for any physically meaningful parameterization. This is proven using Lyapunov functions derived from the physical energy of the bristle elements.
• Unified Framework: The work systematically unifies linear viscoelastic theory with rate-and-state friction laws, providing a clear physical interpretation of friction parameters (micro-stiffness, micro-damping, relaxation times).
• Control-Oriented Design: The paper demonstrates how the passivity property can be exploited for output-feedback control design in robotic systems, ensuring asymptotic tracking stability.

4. Results

• Mathematical Properties:

  • Existence/Uniqueness: The models admit a unique $C^1$ solution for any continuous velocity input and initial conditions.
  • Steady-State: Both models correctly reproduce arbitrary steady-state friction characteristics (e.g., Stribeck curves).
  • Boundedness & Passivity: Proofs confirm that for bounded inputs, the internal states and friction forces remain bounded. The passivity inequality $\int p f(t) v(t) dt \geq V(x(t)) - V(x_0)$ holds, where $V$ is the stored energy.

• Dynamical Behavior (Simulation):

  • Frictional Lag: The models successfully reproduce hysteresis in the force-velocity relationship. The width of the hysteresis loop varies with the rate of velocity change, matching experimental observations (e.g., Hess and Soom).
  • Relaxation: The models capture stress relaxation behavior. By increasing the order $n$ (number of branches), the models can simulate materials with broad frequency spectra (e.g., polymers), showing transient phases before reaching steady-state friction.

• Control Application:

  • A 1-DOF robotic manipulator example was simulated.
  • Using a friction observer based on the FrBD model and a passivity-based control law, the system achieved asymptotic tracking ($\lim_{t\to\infty} \tilde{q}(t) = 0$).
  • The proof relies on the interconnection of two passive operators: the robot dynamics and the friction model.

5. Significance

This paper represents a significant advancement in friction modeling for control engineering:

1. Stability Assurance: By guaranteeing passivity, the models remove a major hurdle in designing stable controllers for systems with friction, eliminating the need for ad-hoc damping terms often required by LuGre.
2. Material Fidelity: The ability to tune the number of relaxation branches ($n$) allows the model to accurately represent highly viscoelastic materials (rubbers, polymers) used in modern robotics and tires, which single-time-constant models cannot capture.
3. Physical Consistency: The parameters have direct physical meanings (stiffness, damping, relaxation time), facilitating their identification in multibody simulations and experimental validation.
4. Future Impact: The framework provides a solid theoretical foundation for future research into nonlinear rheological extensions and the integration of advanced friction models into complex robotic control architectures.

In summary, the authors have successfully extended the FrBD framework to include general linear viscoelasticity, resulting in a family of friction models that are mathematically robust, physically interpretable, and ideally suited for stability-critical control applications.