Cavallo - CAPESB 2021

X Reunión del Capítulo Español de la Sociedad Europea de Biomecánica (ESB)
25 - 26 de octubre 2021, Granada
Identification of human joint impedance using a wearable powered knee exoskeleton
Cavallo G. 1, Lataire J.1, Font-Llagunes J.M. 2, *Cenciarini M.2

1
Vrije Universiteit Brussel, Belgium, [email protected]
2
Universitat Politècnica de Catalunya, Spain, [email protected]
Introduction integration during balance1,5,6,7. However, one of the

Joint impedance is the mechanical property that drawbacks of these studies is that they require
describes the dynamical relationship between joint expensive and bulky custom designed actuated devices
angle and torque. It provides a description of the which are solely used for tailored system identification
neuromechanical behavior of a joint and it is regulated experiments. Recent studies have shown the potential
according to the surrounding environment to promote a of using wearable actuated devices as alternative
stable interaction with it. Joint impedance has been perturbation devices8,9.
shown to be successfully estimated using system In this work we further investigated the viability and
identification techniques on humans experimentally1. usability of a wearable and lightweight powered knee
Estimation of joint impedance is critical in post-stroke exoskeleton as a perturbation device for estimating
individuals particularly when they are in the chronic knee joint impedance. This work leads the way to
state of the pathology (after six months from onset). develop a reproducible protocol to quantitatively
After this time, the affected limbs commonly show signs estimate human joint impedance, particularly in chronic
of increased resistance to movements. This condition of post-stroke individuals. The proposed method can
altered joint impedance is clinically described as joint provide insightful indications of the underlying
hyper-resistance2. The presence of hyper-resistance neuromuscular causes contributing to the hyper-
provokes pain, restricts the range of motion of the resistance (passive, reflexive, and co-contraction).
affected joints, limits the achievement of functional Preliminary results from a pilot testing are presented.
tasks, and might lead to health complications, not
including worsening of the quality of life.
Joint hyper-resistance can have both mechanical and Materials and Methods
neural causes. Neural causes include muscular Five healthy adult subjects (three female) (age±1SD:
spasticity and increased background activation among 31±7 [yr]) were recruited and screened to ensure no
others, while a mechanical cause is contracture. presence of known neurological, sensory, or muscular
Specific treatments are dependent on the origin of the problems. Data from the male subjects were excluded
impairment, but there is still limited knowledge on the because of technical issues that affected the quality of
pathophysiology of joint hyper-resistance, and, the recordings. The mean anthropometric
moreover, current diagnostic tools are only qualitative. characteristics of the remaining subjects were: height
Currently, the evaluation of hyper-resistance is (169±9 [cm]) and weight (65±18 [kg]).
performed by clinicians using manual perturbations. Subjects participated in a 90-min session including
The relationship between the applied constant-speed informed consenting, anthropometric measurements,
angular perturbations and the corresponding perceived preparation for sEMG recordings, fitting of the
joint reaction torque (i.e., joint resistance) is attributed exoskeleton to the subject’s leg and the perturbation
solely to contracture for slow perturbations, while it is trials.
attributed to both contracture and spasticity for faster All subjects were seated during the data collection. A
movements. The perceived resistance is then custom setup was developed to allow relaxation of the
translated to a numeric value by the clinician to assess muscles responsible for knee flexion and extension
the severity of the impairment. Common metrics include (see Fig. 1A). Position perturbations to the knee joint
the Tardieu scale3 and the (Modified) Ashworth Scale, were applied using a wearable actuator (Fig. 1A and
(M)AS4. Both scores are highly dependent on the 1B).
clinician's perception and experience and provide only Different profiles for the perturbations were tested,
a qualitative estimation of the level of spasticity3. including i) constant-speed and ii) small-amplitude
To address these issues and limitations, we propose a perturbations. The order of the testing conditions was
quantitative approach based on measured joint torque randomized across subjects. To prevent fatigue and
in response to an externally applied joint motion using ensure comfort, resting was allowed between trials.
a motorized and controlled apparatus to perform the In this work preliminary results from the constant-speed
constant speed perturbation similarly to the clinical trials are presented. The perturbation consisted of a
evaluation. Such an approach can provide a ramp&hold movement at constant velocity. Two angular
quantitative and objective estimate of joint resistance. velocities were used: 5 and 20 deg/s, referred to as
System identification methods can be used to analyze “SLOW” and “FAST”, respectively.
the measured input-output signals (angular position and Joint angular position was obtained by an encoder
joint reaction torque) and build a quantitative model of embedded in the exoskeleton system. Knee flexion of
the joint impedance. System identification methods 90 deg and full extension corresponded to a reading of
applied to humans have successfully shown to be a 90 deg and 180 deg, respectively. Knee joint resistance
viable approach to gain more insight about the joint torque was measured using a load cell calibrated to
properties up to how humans perform sensorimotor return knee torque in response to applied joint motion.
X Reunión del Capítulo Español de la Sociedad Europea de Biomecánica (ESB)
25 - 26 de octubre 2021, Granada
Figure 3: Sample data from a subject. A. Knee torque-angle

relationship. B. Estimated stiffness
Discussion and Conclusion

In this work we have set forth to develop and test the
Figure 1: A. A subject seated during the experiments. B. viability of a wearable powered knee exoskeleton to
Actuation system of the wearable knee exoskeleton
estimate the knee joint impedance. The motivation was
The data were downsampled to 100 Hz, sensor spikes to develop a method for a more objective and
artifacts were removed and the contribution of the quantitative estimation of joint resistance as opposed to
weight of the exoskeleton on the torque signal was clinical practice. A simplified joint impedance model
removed. The data were divided into ramp (dynamic) was used to analyze the ramp-and-hold data, proving
and hold (static) segments. The data from the static an intuitive and accessible method for the estimation of
segments were used to estimate the weight of the joint stiffness during conditions mimicking clinical
shank and foot. For each ramp segment and each settings. The estimate shows promising results,
direction of motion (extension and flexion) the ratio capable of suggesting that stiffness of the knee joint
between the measured torque and the knee joint angle changes at different angular positions.
was used to estimate the knee joint stiffness. The preliminary findings from this work support the
viability of such an approach to obtain a quantitative
estimate of joint resistance. Additionally, the analysis of
Results the small-amplitude data is expected to provide a more
dynamically relevant model, providing enough
The human leg for the healthy subjects included in the
information to analyze the impedance over a wide range
study showed a fairly linear behavior in response to the
of frequencies. As last remark, further work including
constant-speed perturbations (see Fig. 2). Torque
more subjects and chronic stroke sufferers is needed to
increased as the joint angle increased because of the
investigate the viability of this method with such
added contribution of gravity as the leg became more
individuals.
extended (i.e. shank close to horizontal orientation).
Funding
The project leading to these results has received funding from
the European Union’s Horizon 2020 research and innovation
programme under Marie Skłodowska-Curie grant agreement
No.801342 (Tecniospring INDUSTRY) and the Government
of Catalonia’s Agency for Business Competitiveness (ACCIÓ)
and from the Research Foundation Flanders (FWO-
Vlaanderen, PhD Fellowship No. 1182019N), the Flemish.
Acknowledgments
We thank M.R. Epke and J. de Miguel Fernández for their
assistance with the development of the system.
Figure 2: Sample data from one subject for a “SLOW” trial.
A sample of torque values plotted versus joint angle is
shown in Fig. 3A. Note that for more flexed joint angle
References
configurations (pink and light-blue color data in Fig. 3A), [1] Cavallo G. et al., Conf. Proc. IEEE Eng. Med. Biol. Soc.
the torque-angle relation is less linear. EMBC, 4835-8, 2020, 2020
[2] van den Noort, J. C. et al., Eur. J. Neurol., 24:981-e38,
The values of the estimated stiffness appear to be the 2017
same independently of the direction of motion [3] Alibiglou L. et al., J Neuroeng Rehabil, 5, 18:1-14, 2008
(flexion/extension) at least for more extended [4] Shorter A.L. et al., IEEE ICORR, 246-51, 2019
configurations of the leg (angles above 120 deg) as [5] Cenciarini M. et al., IEEE TBME, 57(2):267-75, 2010
shown in Fig. 3B. Additionally, what is evident from the [6] Cenciarini M., Peterka R.J., JNP, 95(5):2733-50, 2006
data shown in Fig. 3B and less evident in the joint [7] van der Helm C.T. et al., J. Neurosci. Methods, 119(1):1-
torque-angle relationship (Fig. 3A) is that the stiffness 14, 2002
is angle dependent and it appears to increase with the [8] Tucker M.R. et al., IEEE TBME, 64(10):2331-43, 2017
amount of knee flexion. [9] Nazon Y. et al., J. Biomech. Eng., 21 Jul. 2021 Epub

Cavallo - CAPESB 2021

Uploaded by

Copyright:

Available Formats

Cavallo - CAPESB 2021

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cavallo - CAPESB 2021

Uploaded by

Copyright:

Available Formats

X Reunión del Capítulo Español de la Sociedad Europea de Biomecánica (ESB)

25 - 26 de octubre 2021, Granada

Identification of human joint impedance using a wearable powered knee exoskeleton

Cavallo G. 1, Lataire J.1, Font-Llagunes J.M. 2, *Cenciarini M.2

Introduction integration during balance1,5,6,7. However, one of the

Figure 3: Sample data from a subject. A. Knee torque-angle

Discussion and Conclusion

You might also like