Careful control of joint impedance, or dynamic joint stiffness, is crucial for successful performance of movement. Time-varying system identification (TV-SysID) enables quantification of joint impedance during movement. Several TV-SysID methods exist, but have never been systematically compared. Here, we simulate time-varying joint behavior and propose three performance metrics that enable to quantify and compare TV-SysID methods. Time-varying joint stiffness is simulated using a square wave and subsequently estimated with three TV-SysID methods: the ensemble, short data segment, and basis impulse response function method. These methods were compared based on (1) bias with respect to the simulated joint stiffness, (2) random error across 100 simulation trials, and (3) maximum adaptation speed in joint stiffness that can be captured. This approach revealed that each TV-SysID method has its own unique properties. The simulation method and performance metrics pave the way for developing a framework to quantify the strengths and weaknesses of TV-SysID algorithms for estimating joint impedance.
|Title of host publication||Converging Clinical and Engineering Research on Neurorehabilitation IV|
|Subtitle of host publication||Proceedings of the 5th International Conference on Neurorehabilitation (ICNR2020), October 13–16, 2020|
|Editors||Diego Torricelli, Metin Akay, Jose L. Pons|
|Publication status||Published - 2022|
|Event||ICNR 2020: International Conference on NeuroRehabilitation (Virtual) - |
Duration: 13 Oct 2020 → 16 Oct 2020
|Name||Biosystems and Biorobotics|
|Conference||ICNR 2020: International Conference on NeuroRehabilitation (Virtual)|
|Period||13/10/20 → 16/10/20|
Bibliographical noteGreen Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care
Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.