Estimation of Involuntary Impedance in Multi-joint Arm Movements

Hendrik Börner, Satoshi Endo, Sandra HircheIFAC-PapersOnLine 2019 · DOI

TL;DR

This paper introduces a novel method to estimate "involuntary impedance" – the combined intrinsic and reflexive feedback of the human arm during rapid movements. By applying precisely timed force perturbations and filtering the resulting motion, they can isolate and quantify these crucial, fast-acting human responses, paving the way for safer and more intuitive human-robot collaboration.

What Problem Does This Paper Address?

Physical human-robot interaction (pHRI) is becoming increasingly common in various fields, from manufacturing to healthcare. A critical challenge in pHRI is ensuring stability and safety when humans and robots physically interact. This requires understanding and modeling how humans respond to unexpected events or forces. The human arm's response to external disturbances is complex, involving multiple feedback mechanisms. While voluntary cognitive feedback is important, it's too slow to react to sudden, unpredictable events. In such scenarios, the human nervous system relies on much faster intrinsic (e.g., muscle elasticity) and involuntary reflexive feedback. Existing methods for estimating human arm impedance often struggle to isolate these fast, involuntary responses from slower voluntary ones, or they are limited to simpler movements that don't fully represent real-world pHRI. This paper aims to fill that gap by developing a method to specifically estimate this "involuntary impedance."

Methods

The core of the method involves applying carefully designed external force perturbations to a human participant's arm during a point-to-point movement. These perturbations are shaped to generate a "jerk" (the rate of change of acceleration) whose frequency content is distinct from that of the unperturbed movement. By using a high-pass filter tuned to this frequency difference, the researchers can isolate the "feedback jerk" – the arm's involuntary response.

To ensure that only involuntary feedback is captured, the estimation is performed within a very short time window immediately following the perturbation. This window is deliberately limited to exclude the slower, voluntary cognitive feedback. The impedance components (apparent inertia, damping, and stiffness) are then estimated using a least-squares approach within this short interval. The method is validated first using a simulated neuromechanical model of the human arm and then with experimental data from human participants.

Datasets

The paper utilizes a simulated neuromechanical model of the human arm, as described in Franklin et al. (2007), for validation purposes. For the experimental evaluation, data was collected from five human participants performing point-to-point arm movements. No new benchmark datasets were introduced.

Key Insights and Ideas

  • Isolating the "Fast" Human: The key insight is that by carefully designing the perturbation jerk and using a high-pass filter, one can effectively separate the involuntary, reflexive responses from the slower, voluntary ones. This is crucial for understanding how humans react in unpredictable, dynamic situations.
  • The Importance of Timing: Limiting the estimation window to a very short duration (around 150 ms) is critical. This temporal constraint is the primary mechanism for excluding voluntary feedback, which has significantly longer delays.
  • "Involuntary Impedance" as a Concept: The paper defines and quantifies "involuntary impedance" as the combined effect of intrinsic and involuntary reflexive feedback. This provides a useful framework for understanding the immediate, automatic mechanical properties of the human arm during interaction.
  • Perturbation Design Matters: The specific shape and timing of the force perturbation are not arbitrary. They are engineered to create a distinct frequency signature in the arm's response, enabling effective filtering and isolation of the desired feedback.

Key Results

  • Validation of Variational Dynamics: Simulations showed that the estimated variational dynamics (changes in position, velocity, acceleration, and force) had normalized root mean square errors (NRMSEs) below 7.5% for kinematics and below 15% for forces within the estimation interval.
  • Experimental Estimation: The method was successfully applied to experimental data from human participants. The estimated damping and stiffness matrices showed positive diagonal elements and negative off-diagonal elements, consistent with expected biomechanical properties. The stiffness element K22 was significantly larger, attributed to the primary movement axis.
  • Comparison to Existing Studies: The estimated damping and stiffness values were found to be of a similar order of magnitude to those reported in previous studies, though some differences were noted and attributed to factors like estimation interval duration and admittance control parameters.

Limitations and Future Work

  • Inertia Estimation: The paper relies on a priori estimation of the arm's inertia using anthropometric data and inverse dynamics, rather than estimating it directly from the perturbation data.
  • Symmetric Stiffness Assumption: For the estimation interval, stiffness is assumed to be symmetric. While justified by the dominance of intrinsic stiffness and the small contribution of reflexive stiffness, this is a simplification.
  • Future Work: The authors plan to use this method for the estimation and analysis of involuntary impedance modulation strategies. This suggests future research will explore how these involuntary responses can be intentionally controlled or adapted, potentially for improved human-robot collaboration.

Summary updated

Abstract

Stability analysis in physical human-robot interaction requires consideration of human feedback behavior. In unpredictable scenarios, where voluntary cognitive feedback is too slow to guarantee desired task execution, the central nervous system relies on intrinsic and involuntary reflexive feedback. In this work, we present a method for the estimation of the combined effects of intrinsic and involuntary reflexive feedback in multi-joint arm movements, termed involuntary impedance. We apply external force perturbations that are specifically designed to evoke feedback jerk, which can be isolated by application of a high pass filter, and limit the duration of the estimation interval to guarantee the exclusion of voluntary feedback. The isolation of the feedback behavior is validated in simulation and the estimation of the involuntary impedance components is evaluated in an experiment with human participants.
BibTeX
@inproceedings{borner2019estimation,
  author = {Hendrik Börner and Satoshi Endo and Sandra Hirche},
  title = {{Estimation of Involuntary Impedance in Multi-joint Arm Movements}},
  booktitle = {IFAC-PapersOnLine},
  year = {2019},
  doi = {10.1016/j.ifacol.2019.01.052},
}
Figure