A Complex Stiffness Human Impedance Model With Customizable Exoskeleton Control

Binghan He, Huang Huang, Gray Cortright Thomas, Luis Sentis

Abstract

The natural impedance, or dynamic relationship between force and motion, of a human operator can determine the stability of exoskeletons that use interaction-torque feedback to amplify human strength. While human impedance is typically modelled as a linear system, our experiments on a single-joint exoskeleton testbed involving 10 human subjects show evidence of nonlinear behavior: a low-frequency asymptotic phase for the dynamic stiffness of the human that is different than the expected zero, and an unexpectedly consistent damping ratio as the stiffness and inertia vary. To explain these observations, this article considers a new frequency-domain model of the human joint dynamics featuring complex value stiffness comprising a real stiffness term and a hysteretic damping term. Using a statistical F-test we show that the hysteretic damping term is not only significant but is even more significant than the linear damping term. Further analysis reveals a linear trend linking hysteretic damping and the real part of the stiffness, which allows us to simplify the complex stiffness model down to a 1-parameter system. Then, we introduce and demonstrate a customizable fractional-order controller that exploits this hysteretic damping behavior to improve strength amplification bandwidth while maintaining stability, and explore a tuning approach which ensures that this stability property is robust to muscle co-contraction for each individual.

Introduction

When a human wears a strength-amplifying exoskeleton, the two entities form a single coupled system. To achieve "transparency"—where the robot assists without being felt—controllers often use torque feedback to amplify human movement. However, this feedback loop is notoriously prone to instability, manifesting as vibrations or oscillations.

The core issue lies in the modeling of human impedance (the relationship between force and motion). Traditional models typically employ a Mass-Spring-Damper (MSD) framework, assuming "viscous" damping where resistance is proportional to velocity. However, experimental data reveals that human joints do not adhere to this model. Specifically, human joints exhibit a persistent phase lag of 20-30 degrees even at low frequencies, a phenomenon traditional spring-damper models cannot explain. Furthermore, humans maintain a consistent damping ratio even during muscle co-contraction (stiffening), which contradicts standard linear expectations.

Methods

To address these discrepancies, this research proposes a frequency-domain Complex Stiffness Model.

Modeling and Apparatus

The human joint dynamics are represented using a complex value for stiffness: $S_h(s) = M_h s^2 + K_h(1 + c_h j)$ In this model, $c_h$ represents "hysteretic damping"—damping proportional to displacement but in phase with velocity.

Experimental setup showing the elbow exoskeleton testbed and sensors.

Experiments were conducted using the P0 series elastic elbow joint exoskeleton. Ten healthy subjects (5 male, 5 female) participated in nine distinct perturbation experiments. These involved varying physical loads (4.5 kg), grip strengths (10–27 kg), and bias torques (0–8 Nm) to induce a wide range of muscle co-contraction levels.

Control Design

The authors implemented a Fractional-Order Controller defined by $F(s) = k_f s^{-f}$. Unlike standard PI controllers with fixed 90-degree phase shifts, fractional controllers can be tuned to non-integer orders (e.g., $s^{-0.6}$). This allows the exoskeleton to complement the human's natural phase lag, maximizing power amplification without compromising stability.

Results

Statistical validation using F-tests confirmed that the hysteretic damping term provides a significantly better fit for real human data than traditional linear models.

Frequency response plots for subjects B and F across various experimental conditions.

As shown in Figure 5, the frequency response data highlights the "low-frequency asymptotic phase," confirming the inherent, persistent lag in the human neuromuscular system. Key findings include:

  • Model Accuracy: The hysteretic damping term was found to be more significant than traditional linear damping for describing elbow dynamics.
  • Performance Boost: The fractional-order controller achieved an 81% to 88% improvement in dynamical strength amplification at 10 rad/s compared to state-of-the-art controllers that ignore complex stiffness.
  • Consistency: Identified damping ratios (0.19 to 0.23) were consistent with previous independent studies, validating the complex stiffness approach.

Discussion

The study demonstrates that human joints exhibit hysteretic damping, meaning energy loss per cycle depends on motion amplitude rather than frequency. This explains why humans feel stable even during slow movements. By identifying a power-law relationship in joint stiffness, the authors developed a tuning approach robust to muscle co-contraction, preventing the "chatter" often seen in exoskeleton control.

Limitations and Future Work

While successful for the elbow, weight-bearing joints like the knee and ankle require further validation. Additionally, complex stiffness models are non-causal in the time domain, presenting challenges for real-time simulations. Future research will focus on multi-joint systems and investigating the biological origins of this hysteretic behavior, such as muscle spindle feedback or intrinsic tissue properties.

Summary updated

Abstract

The natural impedance, or dynamic relationship between force and motion, of a human operator can determine the stability of exoskeletons that use interaction-torque feedback to amplify human strength. While human impedance is typically modelled as a linear system, our experiments on a single-joint exoskeleton testbed involving 10 human subjects show evidence of nonlinear behavior: a low-frequency asymptotic phase for the dynamic stiffness of the human that is different than the expected zero, and an unexpectedly consistent damping ratio as the stiffness and inertia vary. To explain these observations, this article considers a new frequency-domain model of the human joint dynamics featuring complex value stiffness comprising a real stiffness term and a hysteretic damping term. Using a statistical F-test we show that the hysteretic damping term is not only significant but is even more significant than the linear damping term. Further analysis reveals a linear trend linking hysteretic damping and the real part of the stiffness, which allows us to simplify the complex stiffness model down to a 1-parameter system. Then, we introduce and demonstrate a customizable fractional-order controller that exploits this hysteretic damping behavior to improve strength amplification bandwidth while maintaining stability, and explore a tuning approach which ensures that this stability property is robust to muscle co-contraction for each individual.
BibTeX
@article{he2020complex,
  author = {Binghan He and Huang Huang and Gray Cortright Thomas and Luis Sentis},
  title = {{A Complex Stiffness Human Impedance Model With Customizable Exoskeleton Control}},
  journal = {IEEE Transactions on Neural Systems and Rehabilitation Engineering},
  year = {2020},
  url = {https://openreview.net/forum?id=5jDRPsNGQI},
}

Loading related content...

Figure